CN118135250B - Image recognition method, device, electronic equipment and medium - Google Patents
Image recognition method, device, electronic equipment and medium Download PDFInfo
- Publication number
- CN118135250B CN118135250B CN202410545950.6A CN202410545950A CN118135250B CN 118135250 B CN118135250 B CN 118135250B CN 202410545950 A CN202410545950 A CN 202410545950A CN 118135250 B CN118135250 B CN 118135250B
- Authority
- CN
- China
- Prior art keywords
- image
- diagonal
- identified
- images
- color channels
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000000605 extraction Methods 0.000 claims description 56
- 230000009466 transformation Effects 0.000 claims description 36
- 239000011159 matrix material Substances 0.000 claims description 34
- 230000002146 bilateral effect Effects 0.000 claims description 31
- 230000002087 whitening effect Effects 0.000 claims description 28
- 238000004590 computer program Methods 0.000 claims description 18
- 239000013598 vector Substances 0.000 claims description 16
- 230000000903 blocking effect Effects 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 5
- 238000012545 processing Methods 0.000 abstract description 6
- 238000012360 testing method Methods 0.000 description 31
- 238000012549 training Methods 0.000 description 18
- 238000010586 diagram Methods 0.000 description 12
- 230000008569 process Effects 0.000 description 9
- 230000014509 gene expression Effects 0.000 description 7
- 230000008859 change Effects 0.000 description 5
- 230000009467 reduction Effects 0.000 description 5
- 238000001514 detection method Methods 0.000 description 4
- 238000005286 illumination Methods 0.000 description 4
- 230000004044 response Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000012417 linear regression Methods 0.000 description 3
- 244000141359 Malus pumila Species 0.000 description 2
- 235000021016 apples Nutrition 0.000 description 2
- 230000003190 augmentative effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000008909 emotion recognition Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000000750 progressive effect Effects 0.000 description 2
- 240000008790 Musa x paradisiaca Species 0.000 description 1
- 235000018290 Musa x paradisiaca Nutrition 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/42—Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Image Analysis (AREA)
Abstract
The application is applicable to the technical field of image processing, and provides an image identification method, an image identification device, electronic equipment and a medium, wherein the method comprises the following steps: acquiring a to-be-identified image set containing a plurality of to-be-identified images; extracting target image characteristics of each image to be identified in a plurality of color channels in a mode of cascade connection of a plurality of layers, wherein the plurality of layers comprise at least two of the following: local image hierarchy, global image hierarchy, inter-image hierarchy; performing feature recognition on the target image features of each image to be recognized in a plurality of color channels to obtain channel recognition results of each image to be recognized in a plurality of color channels; and determining a target recognition result of each image to be recognized according to the channel recognition result of each image to be recognized corresponding to each of the plurality of color channels. According to the method, the target image features of each image to be identified in a plurality of color channels are extracted in a mode of cascade connection of a plurality of layers, image feature redundancy can be reduced from the plurality of layers, and accuracy of image identification is guaranteed.
Description
Technical Field
The present application relates to the field of image processing technologies, and in particular, to an image recognition method, an image recognition device, an electronic device, and a medium.
Background
Color images are widely used in the fields of image object detection, recognition, segmentation, and the like because of the rich information contained therein.
On one hand, the current image recognition method generally utilizes the original pixel value of a color image to recognize, and is easily influenced by external factors such as illumination change, shielding, expression change (especially for a face image) and the like; on the other hand, the color image is usually converted into a gray image for processing, and the image channel information is lost, so that the image recognition accuracy is lower.
Disclosure of Invention
The embodiment of the application provides an image recognition method, an image recognition device, electronic equipment and a medium, which can improve the image recognition precision.
In a first aspect, an embodiment of the present application provides an image recognition method, including:
acquiring a to-be-identified image set containing a plurality of to-be-identified images, wherein each to-be-identified image contains a plurality of color channels;
extracting target image characteristics of each image to be identified in a plurality of color channels in a mode of cascade connection of a plurality of layers, wherein the plurality of layers comprise at least two of the following: local image hierarchy, global image hierarchy, inter-image hierarchy;
Performing feature recognition on the target image features of each image to be recognized in a plurality of color channels to obtain channel recognition results of each image to be recognized in a plurality of color channels;
And determining a target recognition result of each image to be recognized according to the channel recognition result of each image to be recognized corresponding to each of the plurality of color channels.
In a possible implementation manner of the first aspect, a plurality of levels are cascaded, so that levels of feature extraction can be adjusted or added according to actual requirements, so as to obtain target image features of each image to be identified in a plurality of color channels.
For example, the features of the multiple images to be identified can be sequentially extracted according to the mode of firstly local image hierarchy and then global image hierarchy, and finally the hierarchy among the images, so as to finally obtain the target image features of each image to be identified in multiple color channels; the features of the images to be identified can be extracted sequentially in a mode of firstly carrying out image hierarchy and then carrying out global image hierarchy and finally carrying out local image hierarchy, so that the target image features of each image to be identified in a plurality of color channels can be obtained.
In a second aspect, an embodiment of the present application provides an image recognition apparatus, including:
the acquisition module is used for acquiring a to-be-identified image set containing a plurality of to-be-identified images, wherein each to-be-identified image contains a plurality of color channels;
the feature extraction module is used for extracting target image features of each image to be identified in a plurality of color channels in a mode of cascade connection of a plurality of layers, and the plurality of layers comprise at least two of the following: local image hierarchy, global image hierarchy, inter-image hierarchy;
the characteristic recognition module is used for carrying out characteristic recognition on the characteristic of each target image of each image to be recognized in a plurality of color channels to obtain a channel recognition result of each image to be recognized in a plurality of color channels;
the determining module is used for determining the target recognition result of each image to be recognized according to the channel recognition result of each image to be recognized corresponding to each of the plurality of color channels.
In a third aspect, an embodiment of the present application provides an electronic device, including: a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the method according to any one of the first aspects when executing the computer program.
In a fourth aspect, embodiments of the present application provide a computer-readable storage medium storing a computer program which, when executed by a processor, implements a method according to any of the first aspects.
In a fifth aspect, embodiments of the present application provide a computer program product for, when run on an electronic device, causing the electronic device to perform the method of any one of the first aspects.
It will be appreciated that the advantages of the second to fifth aspects may be found in the relevant description of the first aspect, and are not described here again.
Compared with the prior art, the embodiment of the application has the beneficial effects that:
Through integrating the diagonal transformation technology of the image, rows (columns) in the transformed diagonal image integrate information of columns and rows in the original image at the same time, and more rows and column information are integrated, so that the local block information of the image is richer, and the reduction of redundancy among pixels is realized by the subsequent feature extraction operation; meanwhile, through adopting bilateral Two-dimensional whitening reconstruction operations (Two-Dimensional Whitening Reconstruction, TWR) of a local image block level, a global image level and a training sample level, progressive pixel redundancy reduction from a single image local small area to a single image global area and then to three levels among sample set images is realized, meanwhile, inherent texture-like characteristics in the images are reserved, so that the color images have stronger robustness under noise conditions such as illumination, expression change and shielding, and the color image recognition precision is further improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments or the description of the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of an image recognition method according to an embodiment of the present application;
FIG. 2 is a flowchart of an image recognition method according to another embodiment of the present application;
FIG. 3 is a schematic diagram of a diagonal transformation provided by an embodiment of the present application;
FIG. 4 is a schematic diagram of another diagonal transformation provided by an embodiment of the present application;
FIG. 5 is a schematic diagram of local feature extraction according to an embodiment of the present application;
FIG. 6 is a schematic diagram of another image recognition method according to an embodiment of the present application;
fig. 7 is a block diagram of an image recognition apparatus according to an embodiment of the present application;
fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
In the following description, for purposes of explanation and not limitation, specific details are set forth such as the particular system architecture, techniques, etc., in order to provide a thorough understanding of the embodiments of the present application. It will be apparent, however, to one skilled in the art that the present application may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.
It should be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It should also be understood that the term "and/or" as used in the present specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes such combinations.
As used in the present description and the appended claims, the term "if" may be interpreted as "when..once" or "in response to a determination" or "in response to detection" depending on the context. Similarly, the phrase "if a determination" or "if a [ described condition or event ] is detected" may be interpreted in the context of meaning "upon determination" or "in response to determination" or "upon detection of a [ described condition or event ]" or "in response to detection of a [ described condition or event ]".
Furthermore, the terms "first," "second," "third," and the like in the description of the present specification and in the appended claims, are used for distinguishing between descriptions and not necessarily for indicating or implying a relative importance.
Reference in the specification to "one embodiment" or "some embodiments" or the like means that a particular feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of the application. Thus, appearances of the phrases "in one embodiment," "in some embodiments," "in other embodiments," and the like in the specification are not necessarily all referring to the same embodiment, but mean "one or more but not all embodiments" unless expressly specified otherwise. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless expressly specified otherwise.
The image recognition method provided by the embodiment of the application can be applied to electronic equipment such as mobile phones, tablet computers, wearable equipment, vehicle-mounted equipment, augmented reality (augmented reality, AR)/Virtual Reality (VR) equipment, notebook computers, ultra-mobile personal computer (UMPC), netbooks, personal digital assistants (personal DIGITAL ASSISTANT, PDA) and the like, and the embodiment of the application does not limit the specific type of the electronic equipment.
Fig. 1 is a schematic flow chart of an image recognition method according to an embodiment of the present application, which can be applied to an electronic device by way of example and not limitation.
It is considered that, when image feature extraction is performed by the conventional image recognition method, only the two-dimensional whitening reconstruction operation is generally performed on the global image or the local block, and other channel information of the image is easily lost, resulting in lower image recognition accuracy. Therefore, aiming at the color image, how to reduce redundancy among features, and improve the robustness and the recognition accuracy of the color image under the noise conditions of illumination, expression change, shielding and the like is a technical problem which is needed to be solved by the person skilled in the art.
Based on the above, the embodiment of the application provides an image recognition method, which ensures that the local block information of the image is richer by fusing more rows and columns of information, and is beneficial to the realization of the reduction of redundancy among pixels in the subsequent feature extraction operation; meanwhile, through adopting the bilateral two-dimensional whitening reconstruction operation of the cascade local image block level, the global image level and the training sample level, progressive pixel redundancy reduction from a single image local small area to a single image global area and then to three levels among sample set images is realized, so that the color image has stronger robustness under the noise conditions of illumination, expression change, shielding and the like, and the color image recognition precision is further improved. As shown in fig. 1, the method includes:
S101, acquiring a to-be-identified image set containing a plurality of to-be-identified images, wherein each to-be-identified image contains a plurality of color channels.
The image set to be identified may be considered as an image set to be identified, where the image set to be identified includes a plurality of images to be identified, and each image to be identified may include a plurality of color channels, and in this embodiment, the type and size of the image to be identified are not limited, for example, the image to be identified may be a color image, and may be represented in RGB, YUV, HSV, or other formats.
S102, extracting target image characteristics of each image to be identified in a plurality of color channels in a mode of cascade connection of a plurality of layers, wherein the plurality of layers comprise at least two of the following: local image hierarchy, global image hierarchy, and inter-image hierarchy.
The process of extracting features of the target image by using a manner of cascade connection of multiple layers may be understood as sequentially extracting features of the image to be identified according to different layers, so as to obtain final features of the target image, where the multiple layers may include at least two of the following: local image hierarchy, global image hierarchy, and inter-image hierarchy. The feature extraction of the local image hierarchy may refer to local feature extraction of the image to be identified, for example, corresponding operations may be performed on image blocks local to the image to be identified to obtain a local feature image of the image to be identified; the feature extraction of the global image hierarchy may refer to global feature extraction of an image to be identified, for example, the global feature image of the image to be identified may be obtained by performing corresponding operation on the whole of the image to be identified; the feature extraction performed by the inter-image hierarchy may refer to performing inter-image feature extraction on a plurality of images to be identified, for example, performing corresponding operations on all the images to be identified according to different color channels to obtain feature images of corresponding color channels, and the like.
Specifically, a plurality of hierarchical cascading modes can be adopted to extract target image features of each image to be identified in a plurality of color channels, the specific process of extracting the target image features is not limited in this embodiment, and exemplary, the features of a plurality of images to be identified can be sequentially extracted according to a mode of firstly local image hierarchy and then global image hierarchy and finally inter-image hierarchy, and finally the target image features of each image to be identified in a plurality of color channels can be obtained; the characteristics of a plurality of images to be identified can be extracted in sequence according to the mode of firstly carrying out image hierarchy and then carrying out global image hierarchy and finally carrying out local image hierarchy so as to obtain the target image characteristics of each image to be identified in a plurality of color channels; the level of feature extraction can be adjusted or reduced according to actual requirements to obtain target image features of each image to be identified in a plurality of color channels.
S103, performing feature recognition on the target image features of each image to be recognized in a plurality of color channels to obtain channel recognition results of each image to be recognized in a plurality of color channels.
The channel recognition result may be a result obtained by performing feature recognition on the target image features of each image to be recognized in the plurality of color channels.
After obtaining the target image characteristics of each image to be identified in a plurality of color channels, performing characteristic identification on the target image characteristics of each image to be identified in a plurality of color channels to obtain channel identification results of each image to be identified corresponding to each color channel, wherein the specific characteristic identification content can be determined according to actual requirements, and the characteristic identification content can be identity identification, for example, the face identification can be performed on the target image characteristics corresponding to each color channel to determine that the image is a person, namely the channel identification results of the respective color channels; the content of the feature recognition can also be emotion recognition, for example, emotion recognition can be performed on the target image features corresponding to each color channel to determine whether the images are happy expressions, gas-generating expressions or painful expressions; the content of the feature recognition can also be target classification, for example, the target image feature corresponding to each color channel can be recognized by target classification, so as to obtain an object represented by the image to be recognized, for example, the object is recognized as a fruit, banana, automobile or flower. Or the content of the feature recognition can be similarity search, such as searching from a preset database to obtain a similar target or object.
Taking the image to be identified a containing three color channels and taking the content of feature identification as an example, the feature identification of the image to be identified a on the respective target image features of the three color channels can be: performing target classification identification on target image characteristics of the image a to be identified in the first color channel to obtain a channel identification result, such as apples, corresponding to the image a to be identified in the first color channel; performing target classification identification on target image characteristics of the image a to be identified in the second color channel to obtain a channel identification result, such as apples, of the image a to be identified corresponding to the second color channel; and carrying out target classification identification on target image characteristics of the image a to be identified in the third color channel to obtain a channel identification result, such as orange, corresponding to the image a to be identified in the third color channel.
The specific recognition means may, for example, utilize a linear regression classifier to calculate the target image features of the image a to be recognized in the first color channel and each image feature of the training image set in the first color channel, so as to obtain a channel recognition result corresponding to the image a to be recognized in the first color channel by comparing. The training image set may include a plurality of pre-acquired training images, each of which is not limited in type and size.
S104, determining a target recognition result of each image to be recognized according to the channel recognition results of each image to be recognized corresponding to the color channels.
The target recognition result may refer to a recognition result finally determined by the image to be recognized.
After the channel recognition results of each image to be recognized in the plurality of color channels are obtained through the steps, the final recognition result of each image to be recognized can be further determined, and after the three-channel recognition results are obtained by performing face recognition on the target image features of each image to be recognized in the single color channel by using the linear regression classifier, the final recognition of the image to be recognized can be realized by using the voting method on the three-channel recognition results, for example, when the channel recognition results of the image a to be recognized in the three color channels are respectively 'Zhang san', 'Li Si', the target recognition result of the image to be recognized can be obtained by using the voting method.
According to the image recognition method provided by the embodiment, a to-be-recognized image set containing a plurality of to-be-recognized images is obtained, and each to-be-recognized image contains a plurality of color channels; extracting target image characteristics of each image to be identified in a plurality of color channels in a mode of cascade connection of a plurality of layers, wherein the plurality of layers comprise at least two of the following: local image hierarchy, global image hierarchy, inter-image hierarchy; performing feature recognition on the target image features of each image to be recognized in a plurality of color channels to obtain channel recognition results of each image to be recognized in a plurality of color channels; and determining a target recognition result of each image to be recognized according to the channel recognition result of each image to be recognized corresponding to each of the plurality of color channels. By utilizing the method, the target image characteristics of each image to be identified in a plurality of color channels are extracted in a mode of cascade connection of a plurality of layers, wherein the plurality of layers comprise at least two of the following: the local image level, the global image level and the inter-image level can reduce redundancy of image features from multiple levels, and accuracy of subsequent image recognition is guaranteed.
Fig. 2 is a flow chart of an image recognition method according to another embodiment of the present application, where extracting target image features of each image to be recognized in a plurality of color channels by using a plurality of hierarchical cascading modes is further optimized as follows: respectively constructing diagonal images of each image to be identified in a plurality of color channels, wherein the diagonal images are images obtained by performing diagonal transformation on the channel images of each image to be identified in the plurality of color channels; performing feature extraction on diagonal images of each image to be identified in a plurality of color channels in a multi-level cascading mode to obtain target image features of each image to be identified in a plurality of color channels, wherein the target image features at least comprise: bilateral two-dimensional whitening features. As shown in fig. 2, the method includes:
S201, acquiring a to-be-identified image set containing a plurality of to-be-identified images, wherein each to-be-identified image contains a plurality of color channels.
S202, respectively constructing diagonal line images of each image to be identified in a plurality of color channels, wherein the diagonal line images are images obtained by performing diagonal line transformation on channel images of each image to be identified in a plurality of color channels.
The diagonal image may be an image obtained by performing diagonal transformation on channel images of each image to be identified in a plurality of color channels, and the channel images may be images corresponding to the image to be identified in a certain color channel.
In this embodiment, diagonal images of each image to be identified in a plurality of color channels may be respectively constructed to perform subsequent feature extraction, where the specific steps of constructing the diagonal images are not limited, for example, the diagonal images of the image to be identified in each color channel may be directly obtained by inputting the channel images of the image to be identified in each color channel into a transformation model, and the transformation model may be used to perform diagonal transformation on the input images; the diagonal line image corresponding to the image to be identified can also be obtained by performing corresponding calculation on the channel image of the image to be identified, and specific calculation steps are not limited herein, for example, different channel images can correspond to different calculation steps.
In some embodiments, the separately constructing diagonal images of each image to be identified in a plurality of color channels includes:
Splitting channels of each image to be identified to obtain a plurality of channel images corresponding to each image to be identified;
and respectively carrying out diagonal transformation on each channel image according to the number of rows and the number of columns in each channel image to obtain diagonal images of each image to be identified in a plurality of color channels.
In a specific embodiment, the channel splitting may be performed on each image to be identified to obtain a plurality of channel images corresponding to each image to be identified, for example, when the image to be identified is a color image including three color channels, the channel splitting may be performed to obtain a channel image of the color image in each color channel, so as to obtain three channel images. And then, in the process of constructing diagonal images, diagonal transformation can be respectively carried out on each channel image according to the difference of each channel image, so that the diagonal images of each image to be identified in a plurality of color channels are obtained.
In this embodiment, the diagonal transformation of different channel images may be implemented based on the difference of the number of rows and the number of columns, and, for example, when a is one channel image corresponding to a certain image to be identified, the diagonal image corresponding to the channel image a is denoted by B, and the specific diagonal transformation process may be illustrated by the following two drawings.
Fig. 3 is a schematic diagram of diagonal transformation according to an embodiment of the present application, as shown in fig. 3, when the number of rows (i.e. 4) of the channel image a is smaller than the number of columns (i.e. 5), two identical channel images a may be connected horizontally in a graphical manner, and diagonal images B corresponding to the channel image a formed by the same size image as the channel image a may be taken from the connected matrix by using diagonal lines as boundaries.
Fig. 4 is a schematic diagram of another diagonal transformation provided in an embodiment of the present application, as shown in fig. 4, when the number of rows (i.e. 5) of the channel image a is greater than the number of columns (i.e. 4), two identical channel images a may be vertically connected in a schematic manner, and diagonal images B corresponding to the channel image a formed by the same size image as the channel image a may be taken from the connected matrix by using diagonal lines as boundaries.
In some embodiments, the diagonal transformation may be performed multiple times, that is, the channel image a may obtain the image B through the diagonal transformation, the image B may obtain the image C through the diagonal transformation, and so on, and the specific times may be operated according to the actual requirements. On the basis, more adjacent row and column information can be fused, and an information basis is provided for the follow-up accurate feature recognition.
S203, performing feature extraction on diagonal images of each image to be identified in a plurality of color channels in a multi-level cascading mode to obtain target image features of each image to be identified in a plurality of color channels, wherein the target image features at least comprise: bilateral two-dimensional whitening features.
After constructing and obtaining diagonal images of each image to be identified in a plurality of color channels, the step can adopt a plurality of hierarchical cascading modes to extract characteristics of the diagonal images of each image to be identified in a plurality of color channels, so as to obtain target image characteristics of each image to be identified in a plurality of color channels, wherein the target image characteristics can at least comprise bilateral two-dimensional whitening characteristics, for example, the target image characteristics can be obtained by carrying out characteristic extraction through bilateral two-dimensional whitening reconstruction operation.
In some embodiments, the feature extraction of the diagonal image of each image to be identified in a plurality of color channels by using a plurality of hierarchical cascading modes to obtain the target image feature of each image to be identified in a plurality of color channels includes:
extracting local features of each diagonal image respectively to obtain local feature images corresponding to each diagonal image;
Global feature extraction is carried out on the local feature image of each diagonal image, so that a global feature image corresponding to each diagonal image is obtained;
And for each color channel, carrying out inter-image feature extraction on a plurality of global feature images corresponding to the color channels to obtain target image features of each image to be identified in the color channels.
As a possible implementation manner, first, for each diagonal image of each image to be identified in any color channel, local feature extraction is performed on each diagonal image to obtain a local feature image corresponding to each diagonal image; then, global feature extraction can be carried out on the local feature image obtained by each diagonal image according to the global image hierarchy, so that a global feature image corresponding to each diagonal image is obtained; finally, extracting features among a plurality of global feature images from the aspect of inter-image hierarchy, for example, extracting features among a plurality of global feature images corresponding to the same color channel according to different color channels, so as to obtain target image features of each image to be identified in the color channel; and the method can also carry out inter-image feature extraction and the like on a plurality of global feature images of the same image to be identified according to different images to be identified. The specific steps of local feature extraction, global feature extraction and inter-image feature extraction are not further developed in the embodiment, so long as the corresponding feature extraction can be realized.
S204, performing feature recognition on the target image features of each image to be recognized in the plurality of color channels to obtain channel recognition results of each image to be recognized in the plurality of color channels.
S205, determining a target recognition result of each image to be recognized according to the channel recognition results of each image to be recognized corresponding to the plurality of color channels.
According to the image recognition method provided by the embodiment, a to-be-recognized image set containing a plurality of to-be-recognized images is obtained, and each to-be-recognized image contains a plurality of color channels; respectively constructing diagonal images of each image to be identified in a plurality of color channels, wherein the diagonal images are images obtained by performing diagonal transformation on the channel images of each image to be identified in the plurality of color channels; performing feature extraction on diagonal images of each image to be identified in a plurality of color channels in a multi-level cascading mode to obtain target image features of each image to be identified in a plurality of color channels, wherein the target image features at least comprise: bilateral two-dimensional whitening features; performing feature recognition on the target image features of each image to be recognized in a plurality of color channels to obtain channel recognition results of each image to be recognized in a plurality of color channels; and determining a target recognition result of each image to be recognized according to the channel recognition result of each image to be recognized corresponding to each of the plurality of color channels. By using the method, the diagonal images of each image to be identified in a plurality of color channels are respectively constructed, so that the diagonal images can integrate the image information of the rows and the columns of the image to be identified at the same time, the image information of the diagonal images is enriched, and an image foundation is provided for realizing pixel redundancy reduction for subsequent feature extraction.
In some embodiments, the extracting local features of each diagonal image to obtain a local feature image corresponding to each diagonal image includes:
carrying out local blocking on each diagonal image to obtain a plurality of image blocks corresponding to each diagonal image, wherein no overlapping image exists between each image block;
Carrying out bilateral two-dimensional whitening reconstruction operation on each image block of each diagonal image to obtain a plurality of processed image blocks corresponding to each diagonal image;
And recombining the plurality of processed image blocks corresponding to each diagonal image according to the position distribution before the local segmentation to obtain a local characteristic image corresponding to each diagonal image.
The plurality of image blocks may be understood as a plurality of partial images obtained by dividing a diagonal image, and the size of each image block may be the same or different, so long as no overlapping image between each image block is ensured.
In the process of specifically extracting local features of each diagonal image, a local feature image corresponding to each diagonal image can be obtained by performing bilateral two-dimensional whitening reconstruction operation on a plurality of image blocks forming the diagonal image.
FIG. 5 is a schematic diagram of local feature extraction according to an embodiment of the present application, as shown in FIG. 5, for a certain diagonal image, the local feature extraction may be performed to obtain a plurality of image blocks, and each image block is respectively subjected to bilateral TWR operation to obtain a plurality of processed image blocks corresponding to the diagonal image; and finally, recombining the plurality of processed image blocks according to the position distribution before the local segmentation so as to obtain a local characteristic image corresponding to the diagonal image.
In some embodiments, global feature extraction is performed on a local feature image of each diagonal image, so as to obtain a global feature image corresponding to each diagonal image, including:
Carrying out bilateral two-dimensional whitening reconstruction operation on the local feature image of each diagonal image to obtain an operated feature image corresponding to each diagonal image;
and performing anti-diagonal transformation operation on the operated characteristic images corresponding to each diagonal image to obtain a global characteristic image corresponding to each diagonal image.
In a specific embodiment, global feature extraction can be performed on each local feature image on the basis of local feature extraction to obtain global feature images corresponding to each diagonal image, for example, bilateral two-dimensional whitening reconstruction operation can be performed on the local feature images of each diagonal image to obtain operated feature images corresponding to each diagonal image, where the operated feature images can be understood as bilateral two-dimensional whitening feature images of a global level, and pixel sequence number arrangement is the same as that of the diagonal images, so that further anti-diagonal transformation operation is required to obtain feature images of normal pixel sequence number arrangement, that is, anti-diagonal transformation operation can be performed on the operated feature images corresponding to each diagonal image to obtain global feature images corresponding to each diagonal image, where the pixel sequence number arrangement of the global feature images is the same as that of the images to be identified. The anti-diagonal transformation operation may be regarded as an operation opposite to the above-described diagonal transformation operation.
In some embodiments, the performing feature extraction between images on the plurality of global feature images corresponding to the color channel to obtain a target image feature of each image to be identified in the color channel includes:
Generating a global data matrix corresponding to the color channel based on a plurality of global feature images corresponding to the color channel, wherein the global data matrix consists of image vectors formed by each global feature image;
And carrying out bilateral two-dimensional whitening reconstruction operation on the global data matrix to obtain the target image characteristics of each image to be identified in the color channel.
The global data matrix may be formed by an image vector formed by each global feature image, and the image vector may refer to a vector formed by the global feature images, for example, a vector formed by stitching according to a fixed row or a fixed column, a vector formed by randomly forming elements in the global feature images, or the like, which is not limited in this embodiment.
In the specific process of extracting features between images, a global data matrix corresponding to each color channel can be generated for each color channel, for example, image vectors corresponding to each global feature image under a certain color channel can be combined together to generate the global data matrix corresponding to the color channel. And secondly, carrying out bilateral two-dimensional whitening reconstruction operation on the generated global data matrix, thereby obtaining the target image characteristics of each image to be identified in each color channel.
Taking the recognition of the object classification as the content of the feature recognition as an example, an exemplary description is given below of the image recognition method provided in the present embodiment.
Firstly, a standard training image set can be obtained, and a test image set to be classified (i.e. a plurality of images to be recognized) is constructed, wherein the training image set and the test image set can be color images of three channels, such as RGB, YUV, HSV, etc., the different formats can be converted mutually, and the training image set can be expressed as，，，For training the number of images, the test image set may be expressed as，，，In order to test the number of images,AndThe dimensions of the rows and columns in the training image or the test image, respectively, are 3 for both the training image and the test image.
Then, feature extraction can be performed on the standard training image set and the test image set respectively to obtain target image features of each test image and each training image in a plurality of color channels, and classification and identification of each test image can be realized by utilizing the target image features of each training image in a plurality of color channels.
FIG. 6 is a schematic diagram of another image recognition method according to an embodiment of the present application, as shown in FIG. 6, taking an example of extracting image features of a test image set as an example, first, diagonal images of each test image in each color channel can be constructed, i.e. each test image can be obtainedWhen the test image is a three-channel color image and expressed by RGB, the test image can be split on three color channels of red (R), green (G) and blue (B) to obtain three channel images respectively, and each channel image can be recorded as（) (Namely, each image to be identified is split in color channel to obtain a plurality of channel images corresponding to each image to be identified); for each channel image, according to the difference of the number of rows and the number of columns (for example, according to the size of the number of rows and the number of columns), diagonal transformation is performed to obtain a corresponding diagonal image (denoted as B), that is, according to the number of rows and the number of columns in each channel image, diagonal transformation is performed to each channel image, so as to obtain diagonal images of each image to be identified in multiple color channels.
Secondly, extracting bilateral TWR characteristic images of the image local block hierarchy, namely carrying out local blocking on each diagonal image B, sequentially carrying out bilateral TWR operation on each image block, and recombining all the image blocks subjected to bilateral TWR processing into an image according to the original blocking positions after completionL is a bilateral two-dimensional whitened image of a local hierarchy.
In particular, for diagonal imagesThe image B can be divided into a plurality of sizes in a non-overlapping mannerMatrix image block of (a)（) (Namely, carrying out local blocking on each diagonal image to obtain a plurality of image blocks corresponding to each diagonal image, wherein no overlapping image exists between each image block); for each matrix image blockTWR operation can be performed, the result is transposed, TWR operation is performed again, and accordingly bilateral TWR operation on matrix image blocks is achieved (namely bilateral two-dimensional whitening reconstruction operation is performed on each image block of each diagonal image respectively, and a plurality of processed image blocks corresponding to each diagonal image are obtained); and finally, the processing result can be recombined into an image L according to the original block position, the size of the image L is the same as that of the image B (namely, a plurality of processed image blocks corresponding to each diagonal image are recombined according to the position distribution before partial block so as to obtain a partial characteristic image corresponding to each diagonal image).
Wherein TWR operation on a certain matrix image block Y may include the following steps.
(1) The matrix image block Y is in-column centred. Let matrix image blockCalculating the mean vector of the column vectors of the matrix image block QSubtracting the mean value from each column of the matrix image block QObtaining a matrix image blockAnd further can obtain a re-centralized column data matrix。
(2) Array data matrixSVD decomposition is carried out to obtainBefore reservationFeature vectors corresponding to the maximum non-zero singular values are obtained to obtain a two-dimensional whitening feature image。
Furthermore, the two-sided TWR characteristic image of the global image hierarchy can be extracted on the basis of the steps, namely TWR operation can be respectively carried out on the two-sided two-dimensional whitened image L of the local hierarchy obtained in the steps, the obtained result is transposed, TWR operation is carried out again, and the two-sided two-dimensional whitened image F of the global hierarchy can be obtained; and performing anti-diagonal transformation on each image F to obtain an anti-diagonalized image Fh, wherein the anti-diagonalized image Fh is a global feature image corresponding to the diagonal image (namely, performing bilateral two-dimensional whitening reconstruction operation on the local feature image of each diagonal image to obtain an operated feature image corresponding to each diagonal image, and performing anti-diagonal transformation operation on the operated feature image corresponding to each diagonal image to obtain the global feature image corresponding to each diagonal image).
Subsequently, a sample-level hierarchical bilateral TWR feature may be extracted, i.e. for each color channel, each resulting anti-diagonalized image Fh may be vectorized, e.g. the anti-diagonalized image Fh may be pulled Cheng Xiangliang in a fixed row or column mannerWherein, the method comprises the steps of, wherein,Is vector quantityThereby generating a global image level data matrix based on vectors corresponding to a plurality of test images for each color channel，（) Is the firstVectors corresponding to the test images can be used for obtaining a global image level data matrix aiming at the first color channelObtaining a global image level data matrix for a second color channelObtaining a global image level data matrix for a third color channel; And performing TWR operation on each global image level data matrix, transposing the obtained result, and performing TWR operation again, so that the sample level double-sided TWR characteristics can be obtained, each column in the sample level double-sided TWR characteristics represents the target image characteristics of each test image in the affiliated color channel (namely, generating a global data matrix corresponding to the color channel based on a plurality of global characteristic images corresponding to the color channel, wherein the global data matrix consists of image vectors formed by each global characteristic image, and performing double-sided two-dimensional whitening reconstruction operation on the global data matrix to obtain the target image characteristics of each image to be identified in the color channel.
Finally, a Linear Regression Classifier (LRC) can be utilized to identify the target classification of a single color channel of each test image based on the target image characteristics of each training image in a plurality of color channels, and the classification results of three color channels are voted to realize the identification of the target classification of the test image (namely, the characteristic identification of the target image characteristics of each image to be identified in a plurality of color channels is carried out to obtain the channel identification result of each image to be identified in a plurality of color channels, and the target identification result of each image to be identified is determined according to the channel identification result of each image to be identified in a plurality of color channels).
For example, when the target classification is specifically performed on the test image b, the following operations may be performed on the test image b to determine the target recognition result of the test image b, and exemplary, the target image features of the test image b in the first color channel may be calculated with the target image features of the training images in the first color channel to determine the channel recognition result of the test image b in the first color channel, and meanwhile, the target image features of the test image b in the second color channel may be calculated with the target image features of the training images in the second color channel to determine the channel recognition result of the test image b in the second color channel, and the target image features of the test image b in the third color channel may be calculated with the target image features of the training images in the third color channel to determine the channel recognition result of the test image b in the third color channel. The test image b is any one of the test images in the test image set to be classified.
Fig. 7 is a block diagram of an image recognition apparatus according to an embodiment of the present application, corresponding to the image recognition method of the above embodiment, and only the portions related to the embodiment of the present application are shown for convenience of explanation.
Referring to fig. 7, the apparatus includes:
An acquiring module 301, configured to acquire a set of images to be identified including a plurality of images to be identified, where each image to be identified includes a plurality of color channels;
The feature extraction module 302 is configured to extract target image features of each image to be identified in a plurality of color channels in a manner of cascade of a plurality of layers, where the plurality of layers includes at least two of the following: local image hierarchy, global image hierarchy, inter-image hierarchy;
the feature recognition module 303 is configured to perform feature recognition on the target image features of each image to be recognized in the multiple color channels, so as to obtain channel recognition results corresponding to each image to be recognized in the multiple color channels;
The determining module 304 is configured to determine a target recognition result of each image to be recognized according to the channel recognition results of each image to be recognized corresponding to the plurality of color channels.
According to the image recognition device provided by the embodiment, an acquisition module acquires a to-be-recognized image set containing a plurality of to-be-recognized images, wherein each to-be-recognized image contains a plurality of color channels; extracting target image characteristics of each image to be identified in a plurality of color channels by a characteristic extraction module in a mode of cascade connection of a plurality of layers, wherein the plurality of layers comprise at least two of the following: local image hierarchy, global image hierarchy, inter-image hierarchy; performing feature recognition on the target image features of each image to be recognized in a plurality of color channels through a feature recognition module to obtain channel recognition results of each image to be recognized in a plurality of color channels; and determining the target recognition result of each image to be recognized according to the channel recognition result of each image to be recognized corresponding to each of the plurality of color channels by the determining module. By utilizing the device, the target image characteristics of each image to be identified in a plurality of color channels are extracted in a mode of cascade connection of a plurality of layers, wherein the plurality of layers comprise at least two of the following: the local image level, the global image level and the inter-image level can reduce redundancy of image features from multiple levels, and accuracy of subsequent image recognition is guaranteed.
Optionally, the feature extraction module includes:
the construction unit is used for respectively constructing diagonal images of each image to be identified in a plurality of color channels, wherein the diagonal images are images obtained by performing diagonal transformation on the channel images of each image to be identified in the plurality of color channels;
The feature extraction unit is used for extracting features of diagonal images of each image to be identified in a plurality of color channels in a multi-level cascading mode to obtain target image features of each image to be identified in a plurality of color channels, wherein the target image features at least comprise: bilateral two-dimensional whitening features.
Optionally, the construction unit is specifically configured to:
Splitting channels of each image to be identified to obtain a plurality of channel images corresponding to each image to be identified;
and respectively carrying out diagonal transformation on each channel image according to the number of rows and the number of columns in each channel image to obtain diagonal images of each image to be identified in a plurality of color channels.
Optionally, the feature extraction unit includes:
The first extraction subunit is used for extracting local features of each diagonal image respectively to obtain local feature images corresponding to each diagonal image;
The second extraction subunit is used for carrying out global feature extraction on the local feature image of each diagonal image respectively to obtain a global feature image corresponding to each diagonal image;
And the third extraction subunit is used for extracting the features of the images among the global feature images corresponding to the color channels for each color channel to obtain the target image features of each image to be identified in the color channel.
Optionally, the first extraction subunit is specifically configured to:
carrying out local blocking on each diagonal image to obtain a plurality of image blocks corresponding to each diagonal image, wherein no overlapping image exists between each image block;
Carrying out bilateral two-dimensional whitening reconstruction operation on each image block of each diagonal image to obtain a plurality of processed image blocks corresponding to each diagonal image;
And recombining the plurality of processed image blocks corresponding to each diagonal image according to the position distribution before the local segmentation to obtain a local characteristic image corresponding to each diagonal image.
Optionally, the second extraction subunit is specifically configured to:
Carrying out bilateral two-dimensional whitening reconstruction operation on the local feature image of each diagonal image to obtain an operated feature image corresponding to each diagonal image;
and performing anti-diagonal transformation operation on the operated characteristic images corresponding to each diagonal image to obtain a global characteristic image corresponding to each diagonal image.
Optionally, the third extraction subunit is specifically configured to:
Generating a global data matrix corresponding to the color channel based on a plurality of global feature images corresponding to the color channel, wherein the global data matrix consists of image vectors formed by each global feature image;
And carrying out bilateral two-dimensional whitening reconstruction operation on the global data matrix to obtain the target image characteristics of each image to be identified in the color channel.
It should be noted that, because the content of information interaction and execution process between the above devices/units is based on the same concept as the method embodiment of the present application, specific functions and technical effects thereof may be referred to in the method embodiment section, and will not be described herein.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions. The functional units and modules in the embodiment may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit, where the integrated units may be implemented in a form of hardware or a form of a software functional unit. In addition, the specific names of the functional units and modules are only for distinguishing from each other, and are not used for limiting the protection scope of the present application. The specific working process of the units and modules in the above system may refer to the corresponding process in the foregoing method embodiment, which is not described herein again.
An embodiment of the present application further provides an electronic device, and fig. 8 is a schematic structural diagram of the electronic device provided by the embodiment of the present application, as shown in fig. 8, where the electronic device includes: at least one processor 401, a memory 402, an input means 403, an output means 404 and a computer program stored in the memory 402 and executable on the at least one processor 401, the processor 401 implementing the steps in any of the various method embodiments described above when executing the computer program.
The input means 403 may be used to receive entered numeric or character information and to generate key signal inputs related to user settings and function control of the electronic device. The output 404 may include a display device such as a display screen.
The embodiments of the present application also provide a computer readable storage medium storing a computer program which, when executed by the processor 401, implements the steps of the various method embodiments described above.
Embodiments of the present application provide a computer program product which, when run on an electronic device, causes the electronic device to perform steps that may be carried out in the various method embodiments described above.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the present application may implement all or part of the flow of the method of the above embodiment, and may be implemented by a computer program to instruct related hardware, and the computer program may be stored in a computer readable storage medium, where the computer program may implement the steps of the method embodiments described above when executed by the processor 401. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, executable files or in some intermediate form, etc. The computer readable medium may include at least: any entity or device capable of carrying computer program code to a camera device/electronic apparatus, a recording medium, a computer Memory, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), an electrical carrier signal, a telecommunications signal, and a software distribution medium. Such as a U-disk, removable hard disk, magnetic or optical disk, etc. In some jurisdictions, computer readable media may not be electrical carrier signals and telecommunications signals in accordance with legislation and patent practice.
In the foregoing embodiments, the descriptions of the embodiments are emphasized, and in part, not described or illustrated in any particular embodiment, reference is made to the related descriptions of other embodiments.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus/electronic device and method may be implemented in other manners. For example, the apparatus/electronic device embodiments described above are merely illustrative, e.g., the division of modules or units is merely a logical functional division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection via interfaces, devices or units, which may be in electrical, mechanical or other forms.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
The above embodiments are only for illustrating the technical solution of the present application, and are not limiting; although the application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application, and are intended to be included in the scope of the present application.
Claims (10)
1. An image recognition method, comprising:
Acquiring a to-be-identified image set comprising a plurality of to-be-identified images, wherein each to-be-identified image comprises a plurality of color channels;
extracting target image characteristics of each image to be identified in a plurality of color channels in a mode of cascade connection of a plurality of layers, wherein the plurality of layers comprise at least two of the following: the target image features are extracted based on diagonal images of the images to be identified in a plurality of color channels, wherein the diagonal images are images obtained by performing diagonal transformation on channel images of the images to be identified in a plurality of color channels;
Performing feature recognition on the target image features of each image to be recognized in a plurality of color channels to obtain channel recognition results of each image to be recognized in a plurality of color channels;
and determining a target recognition result of each image to be recognized according to the channel recognition result of each image to be recognized corresponding to each of the plurality of color channels.
2. The method for identifying an image according to claim 1, wherein extracting the target image features of each image to be identified in a plurality of color channels in a plurality of hierarchical cascades comprises:
Respectively constructing diagonal line images of each image to be identified in a plurality of color channels, wherein the diagonal line images are images obtained by performing diagonal line transformation on channel images of each image to be identified in a plurality of color channels;
Performing feature extraction on diagonal images of each image to be identified in a plurality of color channels in a multi-level cascading mode to obtain target image features of each image to be identified in a plurality of color channels, wherein the target image features at least comprise: bilateral two-dimensional whitening features.
3. The image recognition method according to claim 2, wherein the respectively constructing diagonal images of each of the images to be recognized in a plurality of color channels includes:
splitting channels of each image to be identified to obtain a plurality of channel images corresponding to each image to be identified;
And respectively carrying out diagonal transformation on each channel image according to the number of rows and the number of columns in each channel image to obtain diagonal images of each image to be identified in a plurality of color channels.
4. The method for identifying an image according to claim 2, wherein the feature extraction of the diagonal image of each image to be identified in a plurality of color channels by using a plurality of hierarchical cascades to obtain the target image feature of each image to be identified in a plurality of color channels comprises:
Extracting local features of each diagonal image respectively to obtain local feature images corresponding to each diagonal image;
Global feature extraction is carried out on the local feature image of each diagonal image, so that a global feature image corresponding to each diagonal image is obtained;
And for each color channel, carrying out inter-image feature extraction on a plurality of global feature images corresponding to the color channel to obtain target image features of each image to be identified in the color channel.
5. The method of claim 4, wherein the performing local feature extraction on each diagonal image to obtain a local feature image corresponding to each diagonal image includes:
Carrying out local blocking on each diagonal image to obtain a plurality of image blocks corresponding to each diagonal image, wherein no overlapping image exists between each image block;
respectively carrying out bilateral two-dimensional whitening reconstruction operation on each image block of each diagonal image to obtain a plurality of processed image blocks corresponding to each diagonal image;
And recombining the plurality of processed image blocks corresponding to each diagonal image according to the position distribution before the local segmentation to obtain a local characteristic image corresponding to each diagonal image.
6. The method of claim 4, wherein the performing global feature extraction on the local feature image of each diagonal image to obtain a global feature image corresponding to each diagonal image includes:
carrying out bilateral two-dimensional whitening reconstruction operation on the local feature image of each diagonal image to obtain an operated feature image corresponding to each diagonal image;
And performing anti-diagonal transformation operation on the operated characteristic images corresponding to each diagonal image to obtain a global characteristic image corresponding to each diagonal image.
7. The method for identifying images according to claim 4, wherein the performing feature extraction between images on the plurality of global feature images corresponding to the color channels to obtain target image features of each image to be identified in the color channels comprises:
Generating a global data matrix corresponding to the color channel based on a plurality of global feature images corresponding to the color channel, wherein the global data matrix consists of image vectors formed by each global feature image;
And carrying out bilateral two-dimensional whitening reconstruction operation on the global data matrix to obtain target image characteristics of each image to be identified in the color channel.
8. An image recognition apparatus, comprising:
The acquisition module is used for acquiring a to-be-identified image set containing a plurality of to-be-identified images, wherein each to-be-identified image contains a plurality of color channels;
The feature extraction module is used for extracting target image features of each image to be identified in a plurality of color channels in a mode of cascade connection of a plurality of layers, and the plurality of layers comprise at least two of the following: the target image features are extracted based on diagonal images of the images to be identified in a plurality of color channels, wherein the diagonal images are images obtained by performing diagonal transformation on channel images of the images to be identified in a plurality of color channels;
the characteristic recognition module is used for carrying out characteristic recognition on the target image characteristics of each image to be recognized in a plurality of color channels to obtain channel recognition results of each image to be recognized in a plurality of color channels;
the determining module is used for determining a target recognition result of each image to be recognized according to the channel recognition results of each image to be recognized in the plurality of color channels.
9. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the image recognition method of any one of claims 1 to 7 when the computer program is executed by the processor.
10. A computer-readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the image recognition method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410545950.6A CN118135250B (en) | 2024-05-06 | 2024-05-06 | Image recognition method, device, electronic equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410545950.6A CN118135250B (en) | 2024-05-06 | 2024-05-06 | Image recognition method, device, electronic equipment and medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118135250A CN118135250A (en) | 2024-06-04 |
| CN118135250B true CN118135250B (en) | 2024-09-20 |
Family
ID=91242997
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410545950.6A Active CN118135250B (en) | 2024-05-06 | 2024-05-06 | Image recognition method, device, electronic equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118135250B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115761356A (en) * | 2022-11-23 | 2023-03-07 | 河南讯飞人工智能科技有限公司 | Image recognition method and device, electronic equipment and storage medium |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8866899B2 (en) * | 2011-06-07 | 2014-10-21 | Photon Dynamics Inc. | Systems and methods for defect detection using a whole raw image |
| CN109684931B (en) * | 2018-11-26 | 2023-07-04 | 上海集成电路研发中心有限公司 | A Face Recognition Method Based on Color Channels |
| CN114937179B (en) * | 2022-07-27 | 2022-12-13 | 深圳市海清数字技术有限公司 | Junk image classification method and device, electronic equipment and storage medium |
| CN116681929A (en) * | 2023-05-22 | 2023-09-01 | 哈尔滨理工大学 | Wheat crop disease image recognition method |
| CN117333928B (en) * | 2023-12-01 | 2024-03-22 | 深圳市宗匠科技有限公司 | Face feature point detection method and device, electronic equipment and storage medium |
-
2024
- 2024-05-06 CN CN202410545950.6A patent/CN118135250B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115761356A (en) * | 2022-11-23 | 2023-03-07 | 河南讯飞人工智能科技有限公司 | Image recognition method and device, electronic equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 基于局部二元模式的电子胃镜图像识别;贺小虎;阳维;程时丹;张素;;计算机工程;20090905(第17期);第2节 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118135250A (en) | 2024-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110147722B (en) | Video processing method, video processing device and terminal equipment | |
| KR20200046181A (en) | Deep-running-based image correction detection system and method for providing non-correction detection service using the same | |
| CN113869282B (en) | Face recognition method, hyper-resolution model training method and related equipment | |
| CN112183249A (en) | Video processing method and device | |
| Dürre et al. | A HOG-based real-time and multi-scale pedestrian detector demonstration system on FPGA | |
| CN113850748A (en) | Evaluation system and method for point cloud quality | |
| Ge et al. | WGI-Net: A weighted group integration network for RGB-D salient object detection | |
| US12112524B2 (en) | Image augmentation method, electronic device and readable storage medium | |
| Jwaid et al. | Study and analysis of copy-move & splicing image forgery detection techniques | |
| CN115601820A (en) | Face fake image detection method, device, terminal and storage medium | |
| CN113657273B (en) | Method, device, electronic equipment and medium for determining commodity information | |
| CN111222558A (en) | Image processing method and storage medium | |
| CN118135250B (en) | Image recognition method, device, electronic equipment and medium | |
| CN115223181A (en) | Text detection-based method and device for recognizing characters of seal of report material | |
| KR20200046182A (en) | Deep-running-based image correction detection system and method for providing non-correction detection service using the same | |
| Li et al. | Perceptual hashing for color images | |
| CN113674313A (en) | Pedestrian tracking method and device, storage medium and electronic equipment | |
| CN115223018B (en) | Cooperative detection method and device for camouflaged objects, electronic equipment and storage medium | |
| CN116645513A (en) | Watermark extraction method, model training method, device, electronic equipment and medium | |
| CN111738213B (en) | Person attribute identification method and device, computer equipment and storage medium | |
| CN109118424B (en) | Image processing memory management method and management system for target detection | |
| Zhu et al. | Forensic detection based on color label and oriented texture feature | |
| KR102018773B1 (en) | Apparatus and method for detecting object candidate of image using Additive Kernel BING | |
| CN118887722B (en) | A deep fake face detection method based on self-information enhanced contrastive learning | |
| CN112183480B (en) | Face recognition method, device, terminal equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |