CN106156699B - Image processing apparatus and image matching method - Google Patents
Image processing apparatus and image matching method Download PDFInfo
- Publication number
- CN106156699B CN106156699B CN201510147484.7A CN201510147484A CN106156699B CN 106156699 B CN106156699 B CN 106156699B CN 201510147484 A CN201510147484 A CN 201510147484A CN 106156699 B CN106156699 B CN 106156699B
- Authority
- CN
- China
- Prior art keywords
- search
- related coefficient
- fisrt feature
- parameter
- intermediate features
- 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 61
- 230000009466 transformation Effects 0.000 claims description 47
- 241000208340 Araliaceae Species 0.000 claims description 7
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims description 7
- 235000003140 Panax quinquefolius Nutrition 0.000 claims description 7
- 235000008434 ginseng Nutrition 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 description 87
- 230000008569 process Effects 0.000 description 32
- 238000006243 chemical reaction Methods 0.000 description 10
- 238000001914 filtration Methods 0.000 description 6
- 230000004048 modification Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 230000001815 facial effect Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1365—Matching; Classification
-
- 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
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Human Computer Interaction (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of image processing apparatus and image matching method, can reduce the calculation amount and amount of storage of images match processing, while guaranteeing the accuracy of images match.Image processing equipment includes converter unit, the first search unit, the second search unit and judging unit.First search parameter is respectively set as multiple predefined parameters by the first search unit, the first intermediate features figure is generated based on set the first search parameter and fisrt feature figure, calculates the first related coefficient between the first intermediate features figure and template image generated.Second search unit is based on multiple first related coefficients corresponding with multiple predefined parameters, determine the value range of the second search parameter, and the second search parameter is set separately in the value range of the second search parameter, the second intermediate features figure is generated based on set the second search parameter and second feature figure, calculates the second related coefficient between the second intermediate features figure and template image generated.
Description
Technical field
The present invention relates to image processing apparatus and image matching method.
Background technique
In the image matching method of fingerprint matching, the face matching of the prior art etc., pass through calculating input image
With the related coefficient of template image, such as when related coefficient is greater than threshold value it can be determined that input picture and template image
Match.In addition, in the related coefficient of calculating input image and template image, in order to cope with caused by positional shift and direction deflection
The case where matching error, needs to calculate the related coefficient in the case where being deviated at various locations with all directions deflection, to count
Calculation amount and amount of storage further increase.
It specifically,, will in order to reduce calculation amount and amount of storage in the related coefficient of calculating input image and template image
It is the relatively little of intermediate image of data volume that input picture, which converts (scaling), and calculates the phase relation of intermediate image and template image
Number.But the reduction of the data volume with intermediate image, matching precision can also decline.
Summary of the invention
The present invention completes in view of the above problems, and its purpose is to provide a kind of image processing apparatus and images match sides
Method, can reduce the calculation amount and amount of storage of images match processing, while guarantee the accuracy of images match.
According to an aspect of the present invention, a kind of image processing apparatus is provided.Described image processing unit includes: that transformation is single
Member converts input picture, generates fisrt feature figure and the data volume second feature figure big relative to fisrt feature figure；The
First search parameter is respectively set as multiple predefined parameters by one search unit, based on the first set search parameter and
One characteristic pattern generates the first intermediate features figure, calculates the between the first intermediate features figure generated and the template image
One related coefficient, wherein first related coefficient is corresponding with predefined parameter；Second search unit is based on and multiple predetermined ginsengs
Corresponding multiple first related coefficients of number, determine the value range of the second search parameter, and in the second search parameter
Second search parameter is set separately in value range, is generated based on set the second search parameter and second feature figure
Second intermediate features figure calculates the second related coefficient between the second intermediate features figure generated and the template image；Sentence
Order member, in the case where calculated second related coefficient meets predetermined condition, is determined as the input picture and the mould
Plate images match, based on all second search parameters set in the value range of the second search parameter and second feature figure
The second related coefficient between all second intermediate features figures and the template image generated is all unsatisfactory for the feelings of predetermined condition
Under condition, it is determined as that the input picture and the template image mismatch.
According to another aspect of the present invention, a kind of image matching method is provided.Described image matching process includes: to input
Image is converted, and fisrt feature figure and the data volume second feature figure big relative to fisrt feature figure are generated；By the first search
Parameter is respectively set as multiple predefined parameters, is generated among first based on set the first search parameter and fisrt feature figure
Characteristic pattern calculates the first related coefficient between the first intermediate features figure generated and the template image, wherein described the
One related coefficient is corresponding with predefined parameter；Based on multiple first related coefficients corresponding with multiple predefined parameters, determine
The value range of second search parameter；Second search parameter, base are set separately in the value range of the second search parameter
The second intermediate features figure is generated in set the second search parameter and second feature figure, is calculated special among generated second
The second related coefficient between sign figure and the template image；The case where calculated second related coefficient meets predetermined condition
Under, it is determined as that the input picture is matched with the template image；Based on being set in the value range of the second search parameter
All second search parameters and second feature the figure all second intermediate features figures and the template image that generate between the
In the case that two related coefficients are all unsatisfactory for predetermined condition, it is determined as that the input picture and the template image mismatch.
It is relatively small to calculate data volume when searching for first time for image processing apparatus and image matching method according to the present invention
Fisrt feature figure and template image related coefficient, and when determining using calculated related coefficient second of search
Thus parameter value range can only calculate the second relatively large spy of data volume when searching for for second within the scope of parameter value
The related coefficient of sign figure and template image.Therefore, image processing apparatus and image matching method through the invention, can
The accuracy of images match is held in and is carrying out accuracy when images match using second feature figure, can reduce simultaneously
Calculation amount.
Detailed description of the invention
Fig. 1 is the functional block diagram for indicating the image processing apparatus of embodiments of the present invention.
Fig. 2 is the flow chart for indicating the image matching method of embodiments of the present invention.
Specific embodiment
In the following, being explained with reference to embodiments of the present invention.Description referring to the drawings is provided, with help to by
The understanding of example embodiment of the invention defined by appended claims and their equivalents.It include help to understand it is various specific
Details, but they can only be counted as illustratively.It therefore, it would be recognized by those skilled in the art that can be to reality described herein
The mode of applying makes various changes and modifications, without departing from scope and spirit of the present invention.Moreover, in order to keep specification clearer
Succinctly, by omission pair it is well known that the detailed description of function and construction.
In the following, illustrating the image processing apparatus of embodiments of the present invention referring to Fig.1.Fig. 1 is to indicate reality of the invention
Apply the functional block diagram of the image processing apparatus of mode.
As shown in Figure 1, image processing apparatus 1 includes converter unit 11, the first search unit 12,13 and of the second search unit
Judging unit 14.Wherein, image processing apparatus 1 be, for example, smart phone, tablet computer, laptop, fingerprint identification device,
The image processing apparatus of face identification device etc., as long as having the ability handled image data.
Converter unit 11 converts input picture, generates fisrt feature figure and data volume is big relative to fisrt feature figure
Second feature figure.
Wherein, input picture can be the image acquired by image processing apparatus 1 itself by acquisition module, be also possible to
From the received image of other devices.In addition, the content about input picture, with field phase applied by image processing apparatus 1
It closes, for example, input picture is fingerprint image, if image processing apparatus if the application of image processing apparatus 1 is to carry out fingerprint recognition
1 application is to carry out recognition of face, then input picture is facial image.
Specifically, converter unit 11 for example carries out wavelet transformation to input picture and reduces transformation, so that it is special to generate second
Then sign figure for example carries out wavelet transformation to input picture and reduces transformation, again to generate fisrt feature figure.Wherein, raw
Diminution when diminution when at fisrt feature figure converts and generate second feature figure converts difference, thus the data of second feature figure
Amount is greater than fisrt feature figure.For example, the input picture of 1024*1024 pixel is carried out wavelet transformation and reduced to become by converter unit 11
It changes, generates the second feature figure of 64*64 pixel, and the input picture of 1024*1024 pixel is subjected to wavelet transformation and diminution
Transformation generates the second feature figure of 32*32 pixel.When generating fisrt feature figure, after converter unit 11 can also be to generation
Second feature legend is as carried out wavelet transformation and reducing transformation, to generate fisrt feature figure.
Preferably, converter unit 11 converts input picture or second feature figure with different transformation parameters, thus
Generate at least two fisrt feature figure corresponding from different transformation parameters.Specifically, converter unit is to input picture or second
Characteristic pattern carries out wavelet transformation and reduces transformation, generates fisrt feature figure.For example, utilizing different angles when carrying out wavelet transformation
The window function of degree carries out wavelet transformation to input picture or second feature figure, diminution transformation is then carried out again, to generate difference
Multiple fisrt feature figures corresponding with different angle.In concrete example, converter unit 11 respectively with 0 degree, 30 degree, 60 degree, 90 degree,
120 degree, 150 degree of window function carry out wavelet transformation to input picture or second feature figure, then carry out diminution transformation again, from
And generate 6 fisrt feature figures corresponding with above-mentioned angle respectively.As a result, in subsequent processing using fisrt feature figure come into
When row first is searched for, the accuracy of the first search can be improved, and then can more accurately set taking for the second search parameter
It is worth range.
Moreover it is preferred that converter unit 11 is filtered input picture, a filtered input picture is generated, and
And filtered input picture is converted, generate fisrt feature figure and data volume second spy big relative to fisrt feature figure
Sign figure.Specifically, converter unit 11 is before the processing for generate fisrt feature figure and second feature figure, to input picture into
The pretreatment of row such as wavelet filtering, to eliminate the noise in original input picture.Thereby, it is possible to eliminate noise pair in advance
Interference in images match treatment process improves the accuracy rate of images match.Then, converter unit 11 utilizes filtered input
Image generates fisrt feature figure and second feature figure.
In above-mentioned conversion process and filtering processing, specific description has been carried out by taking wavelet transformation, wavelet filtering as an example,
But the present invention is not limited thereto, can also carry out other processing such as dct transform, mean filter.As long as after generating
Fisrt feature figure and second feature figure can show the feature of input picture well.
First search parameter is respectively set as multiple predefined parameters by the first search unit 12, is searched based on set first
Rope parameter and fisrt feature figure generate the first intermediate features figure, calculate the first intermediate features figure generated and the Prototype drawing
The first related coefficient as between.Wherein the first related coefficient is corresponding with predefined parameter.
Wherein, predefined parameter for example indicates position offset.In the first search process carried out by the first search unit 12
In, such as the first search parameter is respectively set as to different multiple position offsets.Specifically, position offset is by first
Between characteristic pattern corresponding line number of the central point in fisrt feature figure and columns indicate.Plurality of position offset can be pre-
First it is set as (1 row, 1 column), (1 row, 3 column), (1 row, 5 column) ... (15 rows, 15 column) etc..In addition, above-mentioned preset
Multiple position offsets are an example, can be set as needed other position offsets.In addition, predefined parameter for example may be used
To indicate direction rotation amount.At this point, joining in the first search process carried out by the first search unit 12, such as by the first search
Number is respectively set as different multiple directions rotation amounts.Wherein, multiple directions rotation amount can be redefined for -40 degree, -30
Degree, -20 degree ..., 40 degree etc..In addition, above-mentioned preset multiple directions rotation amount is an example, can according to need
Set other directions rotation amount.In addition, predefined parameter can also for example indicate position offset and direction rotation amount simultaneously, into
And it also can according to need the other parameters of expression.
In the following, subsequent explanation is unfolded so that predefined parameter indicates position offset and direction rotation amount simultaneously as an example.This
When, the first search parameter is each set to the various combination of above-mentioned multiple position offsets and above-mentioned multiple directions rotation amount.
Specifically, it after the first search parameter is set to some position offset and some direction rotation amount, is based on
Set the first search parameter and fisrt feature figure generates the first intermediate features figure, and calculates among generated first
The first related coefficient between characteristic pattern and the template image.At this point, calculated first related coefficient and some position
Offset is related to some direction rotation amount.Then, then the first search parameter is set as position offset and direction rotates
Other combinations of amount, repeat above-mentioned processing, to calculate other groups with the position offset and direction rotation amount
Close relevant first related coefficient.Repeat above-mentioned processing, until position offset and direction rotation amount all combinations all
Once it was set to the first search parameter, it is right respectively with all combinations of position offset and direction rotation amount thus, it is possible to calculate
The first related coefficient answered.
After the first search parameter is set to position offset (a row, b column) and direction rotation amount z degree, the first search
Unit 12 utilizes the based on set the first search parameter (that is, position offset (a row, b column) and direction rotation amount z degree)
One characteristic pattern generates the first intermediate features figure.In specific processing, for example, 16*16 pixel fisrt feature figure by including
The matrix of 16*16 element indicates.
Specifically, it indicates the element of the matrix B of the first intermediate features figure generated and indicates the matrix A of fisrt feature figure
Element between relationship it is as follows.The element of the first row first row of matrix B is the element of the xth row y column of matrix A.Its
In, in the case where z is less than or equal to 0, x=a- (cosz °-sinz °) * d/2, y=b- (cosz °-sinz °) * d/2.In addition,
In the case that z is greater than 0, x=a+ (cosz °+sinz °) * d/2, y=b+ (cosz °-sinz °) * d/2.It is in fisrt feature figure
In the case where 16*16 pixel, d=16.The element of the first row secondary series of matrix B is (x-sinz °) row (y+ of matrix A
Cosz °) column element.The tertial element of the first row of matrix B is (x-2*sinz °) row (y+2* of matrix A
Cosz °) column element.In addition, the element of the second row first row of matrix B is (x+cosz °) row (y+ of matrix A
Sinz °) column element.The element of the third line first row of matrix B is (x+cosz °+cosz °) row (y+ of matrix A
Sinz °+sinz °) column element.The element of second row secondary series of matrix B is (x+cosz °-sinz °) row of matrix A
The element of (y+sinz °+cosz °) column.That is, the element that the line n m of matrix B is arranged is, matrix A (x- (m-1) * sinz °+
(n-1) * cosz °) row (cosz ° of y+ (m-1)+(n-1) * sinz °) column element.It so, it is possible to calculate all of matrix B
The value of element.
In addition, in the processing of the element value of above-mentioned calculating matrix B, such as calculated line number or columns (x- (m-
1) * sinz °+(n-1) * cosz °), in the case that (cosz ° of y+ (m-1)+(n-1) * sinz °) be not integer, it is for example carried out
It rounds up, to obtain the line number and columns of integer value.
Preferably, the first search unit 12 determines the picture in fisrt feature figure based on the first set search parameter
Element, and merely with the value of identified pixel, generate the first intermediate features figure.
Specifically, in the first search unit 12 based on the first set search parameter (that is, position offset (a row, b
Column) and direction rotation amount z degree) when generating the first intermediate features figure using fisrt feature figure, as described above, according to expression first
The member of the matrix A of characteristic pattern usually computational chart show the first intermediate features figure generated matrix B element.In calculating matrix A
In line number or when columns, it may appear that the case where line number or columns beyond matrix A.For example, in the matrix that matrix A is 16*16
In the case where, if calculated line number or columns are more than 16 or less than 1, the value of the corresponding element in matrix B is directly set
For null value.Correspondingly, when the first search parameter is set to specific predefined parameter, matrix B is being generated using matrix A
When, in the presence of the element for the matrix that will not be utilized to generate the first intermediate features figure in the matrix of fisrt feature figure, therefore
In specific treatment process, by not reading the element that will not be utilized to generate the matrix of the first intermediate features figure, in phase
The element will not be utilized in the calculating of relationship number, so as to reduce calculation amount.Due to the element of the matrix of fisrt feature figure
Corresponding to pixel, therefore the first search unit 12 determines the picture in fisrt feature figure based on the first set search parameter
Element, and merely with the value of identified pixel, generate the first intermediate features figure.
In addition, it is above-mentioned based on set the first search parameter (that is, position offset (a row, b column) and direction rotate
Amount z degree) processing of the first intermediate features figure is generated using fisrt feature figure is only an example, it can also adopt with other methods
Generate the first intermediate features figure, such as in order to improve the computational accuracy of the first related coefficient between template image, it can also
To carry out conversion process appropriate.
In addition, as described above, converter unit 11 becomes input picture or second feature figure with different transformation parameters
It changes, so that the first search unit 12 exists in the case where generating at least two fisrt feature figure corresponding from different transformation parameters
When generating the first intermediate features figure, the first intermediate features figure is generated using at least two fisrt feature figures.Specifically, it is assumed that raw
At 6 fisrt feature figures, indicate that the matrix of 6 fisrt feature figures is A1~A6.For example, based on the first set search
When parameter (that is, position offset (a row, b column) and direction rotation amount z degree) Lai Shengcheng the first intermediate features figure, as described above, really
Surely the line number and columns in the matrix of fisrt feature figure are indicated, is then read respectively in 6 matrixes for indicating 6 fisrt feature figures
The value of the element of corresponding row and column is taken, and the value of read 6 elements is for example weighted and averaged, to calculate
Indicate the value of the element of the matrix B of the first intermediate features figure.Specifically, the element of the first row secondary series of matrix B is, to matrix
Member of the element of the (y+cosz °) of row of (x-sinz °) column of A1, (x-sinz °) the (y+cosz °) of row column of matrix A 2
Element, the element of the (y+cosz °) of row of (x-sinz °) column of matrix A 3, matrix A 4 (x-sinz °) row (y+cosz °)
The elements of column, the element of (x-sinz °) the (y+cosz °) of row column of matrix A 5, matrix A 6 (x-sinz °) row (y+
Cosz °) column element be weighted and averaged after value.By calculating the first intermediate features figure and template image that so generate
First related coefficient can be improved the accuracy of the first related coefficient, and then can more accurately set the second search parameter
Value range.
First search unit 12 after generating the first intermediate features figure, calculate the first intermediate features figure generated with
The first related coefficient between the template image.First related coefficient of calculating the first intermediate features figure and template image
Processing, can be carried out using method in the prior art, not be unfolded to be illustrated herein.
Wherein, for the first intermediate features figure generate the first related coefficient template image, preferably with first among
The template image of characteristic pattern same pixel.Thus, it is possible to reduce the calculation amount when calculating the first related coefficient.Furthermore it is preferred that
For the template image for generating the first related coefficient with the first intermediate features figure is, by the way that registered images are carried out and are used for
The identical processing of conversion process for generating fisrt feature figure, to generate the template image.Thereby, it is possible to improve calculated
The confidence level of one related coefficient.
In addition, the above explained processing for generating the first intermediate features figure and calculate the processing of the first related coefficient can also be with
It is parallel to execute.Specifically, after generating some element for indicating the matrix of the first intermediate features figure, the square of the generation is utilized
Member in battle array usually carries out calculating the processing of the first related coefficient.Thereby, it is possible to save needed for calculating the first related coefficient
Time improves the efficiency of images match.
By above-mentioned processing, it is related that the first search unit 12 calculates corresponding to multiple predefined parameters multiple first
Coefficient.Specifically, the first search unit 12 calculates corresponding with all combinations of position offset and direction rotation amount
First related coefficient.
Second search unit 13 is based on determining the second search by calculated multiple first related coefficients of the first search unit
The value range of parameter.
Specifically, the second search unit 13 compares the size of the first related coefficient, and according to the size of the first related coefficient
To determine the value range of the second search parameter in the second search process.Wherein, to the parameter institute of the second search parameter setting
The content of expression is identical as the first search parameter.For example, indicating position offset to the predefined parameter of the first search parameter setting
In the case where the rotation amount of direction, position offset and direction rotation amount are also illustrated that the parameter of the second search parameter setting.
Preferably, in the first related coefficient corresponding with multiple predefined parameters, determining and value maximum first
The corresponding predefined parameter of related coefficient, and it is based on predefined parameter corresponding with maximum first related coefficient of value, determine the
The value range of two search parameters.
For example, in by calculated multiple first related coefficients of the first search unit 12, with position offset (3 rows, 4
Column) and the maximum situation of value of 30 degree of direction rotation amount corresponding first related coefficients under, according to the predefined parameter (3 rows, 4 arrange,
30 degree of direction rotation amount) determine the value range of the second search parameter.Specifically, such as by the value model of the second search parameter
It encloses and is determined as 2-4 row, 3-5 column, direction rotation amount 21-39 degree.
In addition it is also possible to determine the value range of the second search parameter by other methods.For example, determining and value
The maximum corresponding predefined parameter of first related coefficient and predetermined ginseng corresponding with the first second largest related coefficient of value
Number, and based on predefined parameter corresponding to maximum first related coefficient of value and with second largest first related of value
The corresponding predefined parameter of coefficient, determines the value range of the second search parameter.Although the calculation amount meeting of the second search process as a result,
It improves, but correspondingly can be improved the accuracy of images match.
After the value range that the second search parameter has been determined, value of second search unit 13 in the second search parameter
The second search parameter is set separately in range, is generated among second based on set the second search parameter and second feature figure
Characteristic pattern calculates the second related coefficient between the second intermediate features figure generated and the template image.
Specifically, the second search parameter is successively set as some positional shift in value range by the second search unit 13
After amount and some direction rotation amount, the second intermediate features are generated based on set the second search parameter and second feature figure
Figure, and calculate the second related coefficient between the second intermediate features figure generated and the template image.If by aftermentioned
Judging unit 14 processing and be judged as that calculated second related coefficient is unsatisfactory for predetermined condition, then again by second search join
Number is set as other combinations (setting in value range certainly) of position offset and direction rotation amount, repeats above-mentioned place
Reason, to calculate the second related coefficient.
In addition, in the second search process carried out by the second search unit 13, based on the second set search parameter
The processing of the second intermediate features figure is generated with second feature figure and calculates the second intermediate features figure generated and the mould
The processing of the second related coefficient between plate image is identical as the first above-mentioned search process, therefore without repeat description.
In addition, identically as the first search process, the second search unit 13 is determined based on the second set search parameter in the second spy
The pixel in figure is levied, and merely with the value of identified pixel, generates the second intermediate features figure.In addition, at the first search
Reason in the same manner, can also adopt and generate the second intermediate features figure with other methods, such as in order to improve between template image
The first related coefficient computational accuracy, conversion process appropriate can also be carried out.
Wherein, for the second intermediate features figure generate the second related coefficient template image, preferably with second among
The template image of characteristic pattern same pixel.Thus, it is possible to reduce the calculation amount when calculating the second related coefficient.Furthermore it is preferred that
For the template image for generating the second related coefficient with the second intermediate features figure is, by the way that registered images are carried out and are used for
The identical processing of conversion process for generating second feature figure, to generate the template image.Thereby, it is possible to improve calculated
The confidence level of one related coefficient.
In addition, the above explained processing for generating the second intermediate features figure and calculate the processing of the second related coefficient can also be with
It is parallel to execute.Specifically, after generating some element for indicating the matrix of the second intermediate features figure, the square of the generation is utilized
Member in battle array usually carries out calculating the processing of the second related coefficient.Thereby, it is possible to save needed for calculating the second related coefficient
Time improves the efficiency of images match.
Judging unit 14 in the case where calculated second related coefficient meets predetermined condition, be determined as input picture with
Template image matching, based on all second search parameters set in the value range of the second search parameter and second feature
The second related coefficient schemed between all second intermediate features figures and the template image that generate all is unsatisfactory for predetermined condition
In the case of, it is determined as that the input picture and the template image mismatch.
Specifically, after calculating the second related coefficient under some second search parameter by the second search unit 13,
Judge whether to meet predetermined condition by calculated second related coefficient of the second search unit 13 by judging unit 14 (for example,
Whether be greater than threshold value), in the case where second related coefficient meets predetermined condition, judging unit 14 be determined as input picture with
Template image matching.Processing terminate for images match as a result,.When judging unit 14 be determined as it is calculated by the second search unit 13
When second related coefficient is unsatisfactory for predetermined condition, as described above, value model of second search unit 13 in the second search parameter
The second search parameter of interior reset is enclosed, and repeats above-mentioned processing.Second search unit 13 is by the value of the second search parameter
All values in range were all once set to the second search parameter, still not by calculated second related coefficient of the second search unit 13
In the case where meeting predetermined condition, judging unit 14 is it can be determined that input picture and template image mismatch.
The image processing equipment 1 of embodiment according to the present invention passes through the fisrt feature figure relatively small using data volume
The first search process determine the range of the second search process, therefore utilizing the second of the relatively large second feature figure of data volume
It is only calculated in determining range in search process, therefore can reduce the calculation amount in whole image matching treatment, together
When can by images match processing precision be maintained in full scope using data volume it is relatively large second feature figure progress
The level of calculating.
In the following, illustrating the image matching method of embodiments of the present invention referring to Fig. 2.Fig. 2 is to indicate reality of the invention
Apply the flow chart of the image matching method of mode.
Image matching method shown in Fig. 2 can be applied to image processing equipment shown in FIG. 1.As shown in Figure 1, at image
Managing equipment 1 includes converter unit 11, the first search unit 12, the second search unit 13 and judging unit 14.
In step sl, input picture is converted, generates fisrt feature figure and data volume relative to fisrt feature figure
Big second feature figure.
Wherein, input picture can be the image acquired by image processing apparatus 1 itself by acquisition module, be also possible to
From the received image of other devices.In addition, the content about input picture, with field phase applied by image processing apparatus 1
It closes, for example, input picture is fingerprint image, if image processing apparatus if the application of image processing apparatus 1 is to carry out fingerprint recognition
1 application is to carry out recognition of face, then input picture is facial image.
Specifically, converter unit 11 for example carries out wavelet transformation to input picture and reduces transformation, so that it is special to generate second
Then sign figure for example carries out wavelet transformation to input picture and reduces transformation, again to generate fisrt feature figure.Wherein, raw
Diminution when diminution when at fisrt feature figure converts and generate second feature figure converts difference, thus the data of second feature figure
Amount is greater than fisrt feature figure.For example, the input picture of 1024*1024 pixel is carried out wavelet transformation and reduced to become by converter unit 11
It changes, generates the second feature figure of 64*64 pixel, and the input picture of 1024*1024 pixel is subjected to wavelet transformation and diminution
Transformation generates the second feature figure of 32*32 pixel.When generating fisrt feature figure, after converter unit 11 can also be to generation
Second feature legend is as carried out wavelet transformation and reducing transformation, to generate fisrt feature figure.
Preferably, in step sl, input picture is filtered, generates a filtered input picture, and right
Filtered input picture is converted, and fisrt feature figure and the data volume second feature big relative to fisrt feature figure are generated
Figure.Specifically, converter unit 11 carries out input picture before the processing for generate fisrt feature figure and second feature figure
Such as the pretreatment of wavelet filtering, to eliminate the noise in original input picture.Thereby, it is possible to eliminate noise in advance to figure
As the interference during matching treatment, the accuracy rate of images match is improved.Then, converter unit 11 is schemed using filtered input
As generating fisrt feature figure and second feature figure.
Moreover it is preferred that in step sl, being become with different transformation parameters to input picture or second feature figure
It changes, to generate at least two fisrt feature figure corresponding from different transformation parameters.Specifically, 11 pairs of input figures of converter unit
Picture or second feature figure carry out wavelet transformation and reduce transformation, generation fisrt feature figure.For example, when carrying out wavelet transformation, benefit
Wavelet transformation is carried out to input picture or second feature figure with the window function of different angle, then carries out diminution transformation again, thus
Generate multiple fisrt feature figures corresponding with different angle respectively.In concrete example, converter unit 11 is respectively with 0 degree, 30 degree, 60
Then degree, 90 degree, 120 degree, 150 degree of window function reduce input picture or second feature figure progress wavelet transformation again
Transformation, to generate 6 fisrt feature figures corresponding with above-mentioned angle respectively.It is special using first in subsequent processing as a result,
Sign figure when carrying out the first search, can be improved the accuracy of the first search, and then can more accurately set the second search
The value range of parameter.
In above-mentioned conversion process and filtering processing, specific description has been carried out by taking wavelet transformation, wavelet filtering as an example,
But the present invention is not limited thereto, can also carry out other processing such as dct transform, mean filter.As long as after generating
Fisrt feature figure and second feature figure can show the feature of input picture well.
In step s 2, the first search parameter is respectively set as multiple predefined parameters, based on the first set search
Parameter and fisrt feature figure generate the first intermediate features figure, calculate the first intermediate features figure generated and the template image
Between the first related coefficient, wherein first related coefficient is corresponding with predefined parameter.
Wherein, predefined parameter for example indicates position offset and/or direction rotation amount.It is carried out by the first search unit 12
The first search process in, such as the first search parameter is respectively set as different multiple position offsets and/or different
The combination of multiple directions rotation amount.Specifically, position offset is by the central point of the first intermediate features figure in fisrt feature figure
Corresponding line number and columns indicate.Plurality of position offset can be redefined for (1 row, 1 column), (1 row, 3 column), (1
Row, 5 column) ... (15 rows, 15 column) etc..In addition, multiple directions rotation amount can be redefined for -40 degree, -30 degree, -20
Degree ..., 40 degree etc..In addition, above-mentioned preset multiple position offsets and multiple directions rotation amount are an example, it can
Other direction rotation amounts are set as needed.In addition, predefined parameter can also for example indicate position offset and side simultaneously
To rotation amount, and then it also can according to need the other parameters of expression.
Specifically, the first search parameter is being each set to above-mentioned multiple position offsets and the rotation of above-mentioned multiple directions
In the case where the various combination for turning amount, the first search parameter be set to some position offset and some direction rotation amount it
Afterwards, the first intermediate features figure is generated based on set the first search parameter and fisrt feature figure, and calculated generated
The first related coefficient between first intermediate features figure and the template image.At this point, calculated first related coefficient with should
Some position offset is related to some direction rotation amount.Then, then by the first search parameter be set as position offset and
Other combinations of direction rotation amount, repeat above-mentioned processing, to calculate and the position offset and direction rotation amount
Others combine relevant first related coefficient.Above-mentioned processing is repeated, it is all until position offset and direction rotation amount
Combination be all once set to the first search parameter, thus, it is possible to calculate all groups with position offset and direction rotation amount
Close corresponding first related coefficient.
After the first search parameter is set to position offset (a row, b column) and direction rotation amount z degree, the first search
Unit 12 utilizes the based on set the first search parameter (that is, position offset (a row, b column) and direction rotation amount z degree)
One characteristic pattern generates the first intermediate features figure.In specific processing, for example, 16*16 pixel fisrt feature figure by including
The matrix of 16*16 element indicates.
Specifically, it indicates the element of the matrix B of the first intermediate features figure generated and indicates the matrix A of fisrt feature figure
Element between relationship it is as follows.The element of the first row first row of matrix B is the element of the xth row y column of matrix A.Its
In, in the case where z is less than or equal to 0, x=a- (cosz °-sinz °) * d/2, y=b- (cosz °-sinz °) * d/2.In addition,
In the case that z is greater than 0, x=a+ (cosz °+sinz °) * d/2, y=b+ (cosz °-sinz °) * d/2.It is in fisrt feature figure
In the case where 16*16 pixel, d=16.The element of the first row secondary series of matrix B is (x-sinz °) row (y+ of matrix A
Cosz °) column element.The tertial element of the first row of matrix B is (x-2*sinz °) row (y+2* of matrix A
Cosz °) column element.In addition, the element of the second row first row of matrix B is (x+cosz °) row (y+ of matrix A
Sinz °) column element.The element of the third line first row of matrix B is (x+cosz °+cosz °) row (y+ of matrix A
Sinz °+sinz °) column element.The element of second row secondary series of matrix B is (x+cosz °-sinz °) row of matrix A
The element of (y+sinz °+cosz °) column.That is, the element that the line n m of matrix B is arranged is, matrix A (x- (m-1) * sinz °+
(n-1) * cosz °) row (cosz ° of y+ (m-1)+(n-1) * sinz °) column element.It so, it is possible to calculate all of matrix B
The value of element.
In addition, in the processing of the element value of above-mentioned calculating matrix B, such as calculated line number or columns (x- (m-
1) * sinz °+(n-1) * cosz °), in the case that (cosz ° of y+ (m-1)+(n-1) * sinz °) be not integer, it is for example carried out
It rounds up, to obtain the line number and columns of integer value.
In addition, as described above, being become in step sl with different transformation parameters to input picture or second feature figure
It changes, thus in the case where generating corresponding from different transformation parameters at least two fisrt feature figure, generates the in step s 2
When one intermediate features figure, the first intermediate features figure is generated using at least two fisrt feature figures.Specifically, it is assumed that generate 6
A fisrt feature figure indicates that the matrix of 6 fisrt feature figures is A1~A6.For example, based on the first set search parameter
When (that is, position offset (a row, b column) and direction rotation amount z degree) Lai Shengcheng the first intermediate features figure, as described above, determining table
Show the line number and columns in the matrix of fisrt feature figure, the then reading pair respectively in 6 matrixes for indicating 6 fisrt feature figures
The value of the element for the row and column answered, and the value of read 6 elements is for example weighted and averaged, to calculate expression
The value of the element of the matrix B of first intermediate features figure.Specifically, the element of the first row secondary series of matrix B is, to matrix A 1
Element, square of the element of the (y+cosz °) of row of (x-sinz °) column, (x-sinz °) the (y+cosz °) of row column of matrix A 2
Member of the element of the (y+cosz °) of row of (x-sinz °) column of battle array A3, (x-sinz °) the (y+cosz °) of row column of matrix A 4
Element, the element of the (y+cosz °) of row of (x-sinz °) column of matrix A 5, matrix A 6 (x-sinz °) row (y+cosz °)
The element of column be weighted and averaged after value.By the first phase for calculating the first intermediate features figure and template image that so generate
Relationship number can be improved the accuracy of the first related coefficient, and then can more accurately set the value of the second search parameter
Range.
In addition, it is above-mentioned based on set the first search parameter (that is, position offset (a row, b column) and direction rotate
Amount z degree) processing of the first intermediate features figure is generated using fisrt feature figure is only an example, it can also adopt with other methods
Generate the first intermediate features figure, such as in order to improve the computational accuracy of the first related coefficient between template image, it can also
To carry out conversion process appropriate.
Preferably, in step s 2, the pixel in fisrt feature figure is determined based on the first set search parameter, and
And merely with the value of identified pixel, the first intermediate features figure is generated.
Specifically, in step s 2 based on the first set search parameter (that is, position offset (a row, b column) and side
To rotation amount z degree) when generating the first intermediate features figure using fisrt feature figure, as described above, according to fisrt feature figure is indicated
Matrix A member usually computational chart show the first intermediate features figure generated matrix B element.Row in calculating matrix A
When several or columns, it may appear that the case where line number or columns beyond matrix A.For example, the case where matrix A is the matrix of 16*16
Under, if calculated line number or columns are more than 16 or less than 1, the value of the corresponding element in matrix B is directly set as null value.
Correspondingly, when the first search parameter is set to specific predefined parameter, when using matrix A to generate matrix B,
In the presence of the element for the matrix that will not be utilized to generate the first intermediate features figure in the matrix of one characteristic pattern, therefore specifically locating
During reason, by not reading the element that will not be utilized to generate the matrix of the first intermediate features figure, in related coefficient
The element will not be utilized in calculating, so as to reduce calculation amount.Since the element of the matrix of fisrt feature figure corresponds to picture
Element, therefore in step s 2 based on the determining pixel in fisrt feature figure of the first set search parameter, and merely with
The value of identified pixel generates the first intermediate features figure.
In step s 2, after generating the first intermediate features figure, the first intermediate features figure generated and institute are calculated
State the first related coefficient between template image.The place of first related coefficient of calculating the first intermediate features figure and template image
Reason, can be carried out using method in the prior art, not be unfolded to be illustrated herein.Wherein, it is used to and the first intermediate features
Figure generates the template image of the first related coefficient, the preferably template image with the first intermediate features figure same pixel.To energy
Enough calculation amounts reduced when calculating the first related coefficient.Furthermore it is preferred that being to be used to generate the first phase with the first intermediate features figure
The template image of relationship number is, by carrying out place identical with for generating the conversion process of fisrt feature figure to registered images
Reason, to generate the template image.Thereby, it is possible to improve the confidence level of calculated first related coefficient.
In addition, the above explained processing for generating the first intermediate features figure and calculate the processing of the first related coefficient can also be with
It is parallel to execute.Specifically, after generating some element for indicating the matrix of the first intermediate features figure, the square of the generation is utilized
Member in battle array usually carries out calculating the processing of the first related coefficient.Thereby, it is possible to save needed for calculating the first related coefficient
Time improves the efficiency of images match.
By above-mentioned processing, in step s 2, multiple first phase relations corresponding with multiple predefined parameters are calculated
Number (specifically, first related coefficient corresponding with all combinations of position offset and direction rotation amount).
In step s3, multiple first related coefficients corresponding with multiple predefined parameters are based on, determine the second search
The value range of parameter.
Specifically, compare the size of calculated first related coefficient in step s 2 in step s3, and according to first
The size of related coefficient determines the value range of the second search parameter in the second search process.Wherein, the second search is joined
Content represented by the parameter of number setting is identical as the first search parameter.For example, to the predefined parameter of the first search parameter setting
In the case where indicating position offset and direction rotation amount, to the parameter of the second search parameter setting also illustrate that position offset and
Direction rotation amount.
Preferably, in step s3, in the first related coefficient corresponding with multiple predefined parameters, determining and value
The corresponding predefined parameter of maximum first related coefficient, and it is based on predetermined ginseng corresponding with maximum first related coefficient of value
Number, determines the value range of the second search parameter.
For example, in step s 2 in calculated multiple first related coefficients, with position offset (3 rows, 4 column) and direction
In the maximum situation of value of 30 degree of rotation amount corresponding first related coefficients, according to the predefined parameter, (3 rows, 4 column, direction rotate
30 degree of amount) determine the value range of the second search parameter.Specifically, such as by the value range of the second search parameter it is determined as
2-4 row, 3-5 column, direction rotation amount 21-39 degree.
In addition it is also possible to determine the value range of the second search parameter by other methods.For example, determining and value
The maximum corresponding predefined parameter of first related coefficient and predetermined ginseng corresponding with the first second largest related coefficient of value
Number, and based on predefined parameter corresponding to maximum first related coefficient of value and with second largest first related of value
The corresponding predefined parameter of coefficient, determines the value range of the second search parameter.Although the calculation amount meeting of the second search process as a result,
It improves, but correspondingly can be improved the accuracy of images match.
In step s 4, second search parameter is set separately in the value range of the second search parameter, is based on institute
The second search parameter and second feature figure of setting generates the second intermediate features figure, calculates the second intermediate features figure generated
The second related coefficient between the template image.
Specifically, in step s 4, some position offset being successively set as the second search parameter in value range
After some direction rotation amount, the second intermediate features are generated based on set the second search parameter and second feature figure
Figure, and calculate the second related coefficient between the second intermediate features figure generated and the template image.If by aftermentioned
Step S5 processing and be judged as that calculated second related coefficient is unsatisfactory for predetermined condition, then the second search parameter is set again
It is set to other combinations (setting in value range certainly) of position offset and direction rotation amount, repeats above-mentioned processing,
To calculate the second related coefficient.
In addition, in the second search process of step S4, based on set the second search parameter and second feature figure come
Generate the processing of the second intermediate features figure and calculate between the second intermediate features figure generated and the template image the
The processing of two related coefficients is identical as above-mentioned the first search process of step S2, therefore without repeat description.In addition, with
First search process in the same manner, determines the picture in second feature figure based on the second set search parameter in step s 4
Element, and merely with the value of identified pixel, generate the second intermediate features figure.In addition, identically as the first search process,
It can adopt and generate the second intermediate features figure with other methods, such as in order to improve the first phase relation between template image
Several computational accuracies can also carry out conversion process appropriate.
Wherein, for the second intermediate features figure generate the second related coefficient template image, preferably with second among
The template image of characteristic pattern same pixel.Thus, it is possible to reduce the calculation amount when calculating the second related coefficient.Furthermore it is preferred that
For the template image for generating the second related coefficient with the second intermediate features figure is, by the way that registered images are carried out and are used for
The identical processing of conversion process for generating second feature figure, to generate the template image.Thereby, it is possible to improve calculated
The confidence level of one related coefficient.
In addition, the above explained processing for generating the second intermediate features figure and calculate the processing of the second related coefficient can also be with
It is parallel to execute.Specifically, after generating some element for indicating the matrix of the second intermediate features figure, the square of the generation is utilized
Member in battle array usually carries out calculating the processing of the second related coefficient.Thereby, it is possible to save needed for calculating the second related coefficient
Time improves the efficiency of images match.
In step s 5, in the case where calculated second related coefficient meets predetermined condition, it is determined as input picture
It is matched with template image.In addition, in step s 6, based on all second set in the value range of the second search parameter
The second related coefficient between all second intermediate features figures and the template image that search parameter and second feature figure generate
In the case where being all unsatisfactory for predetermined condition, it is determined as that the input picture and the template image mismatch.
Specifically, single by determining after calculating the second related coefficient under some second search parameter in step s 4
Member 14 judgement in step s 4 calculated second related coefficient whether meet predetermined condition (such as, if be greater than threshold value),
In the case where second related coefficient meets predetermined condition, judging unit 14 is determined as that input picture is matched with template image.
Processing terminate for images match as a result,.When judging unit 14 is determined as that calculated second related coefficient is discontented in step s 4
When sufficient predetermined condition, as described above, in step s 4, the second search ginseng is reset in the value range of the second search parameter
Number, and repeat above-mentioned processing.It, in step s 4 will be in the value range of the second search parameter by above-mentioned duplicate processing
All values were all once set to the second search parameter, in the case that calculated second related coefficient of institute is still unsatisfactory for predetermined condition,
Judging unit 14 is it can be determined that input picture and template image mismatch.
The image matching method of embodiment according to the present invention, by utilizing the relatively small fisrt feature figure of data volume
First search process determines the range of the second search process, therefore searches using the second of the relatively large second feature figure of data volume
It is only calculated in determining range in rope processing, therefore can reduce the calculation amount in whole image matching treatment, simultaneously
The precision of images match processing can be maintained in full scope and be counted using the relatively large second feature figure of data volume
The level of calculation.
Those of ordinary skill in the art may be aware that being incorporated in each unit and step of embodiments of the present invention description
Suddenly, it can be realized with electronic hardware, computer software, or a combination of the two.And software module can be placed in arbitrary form
Computer storage medium in.In order to clearly illustrate the interchangeability of hardware and software, in the above description according to function
Each exemplary composition and step can be generally described.These functions are implemented in hardware or software actually, are depended on
In the specific application and design constraint of technical solution.Those skilled in the art can use not each specific application
Described function is realized with method, but such implementation should not be considered as beyond the scope of the present invention.
Each embodiment of the invention has been described in detail above.However, it should be appreciated by those skilled in the art that not
In the case where being detached from the principle and spirit of the invention, these embodiments can be carry out various modifications, combination or sub-portfolio, and
Such modification should be fallen within the scope of the present invention.
Claims (10)
1. a kind of image matching method, comprising:
Input picture is converted, fisrt feature figure and the data volume second feature figure big relative to fisrt feature figure are generated；
First search parameter is respectively set as multiple predefined parameters, based on set the first search parameter and fisrt feature figure
It generates the first intermediate features figure, calculates the first related coefficient between the first intermediate features figure and template image generated,
Wherein first related coefficient is corresponding with predefined parameter；
Based on multiple first related coefficients corresponding with multiple predefined parameters, the value range of the second search parameter is determined；
Second search parameter is set separately in the value range of the second search parameter, based on the second set search ginseng
Several and second feature figure generates the second intermediate features figure, calculate the second intermediate features figure generated and the template image it
Between the second related coefficient；
In the case where calculated second related coefficient meets predetermined condition, it is determined as the input picture and the Prototype drawing
As matching；
What is generated based on all second search parameters set in the value range of the second search parameter and second feature figure
In the case that the second related coefficient between all second intermediate features figures and the template image is all unsatisfactory for predetermined condition,
It is determined as that the input picture and the template image mismatch.
2. image matching method as described in claim 1, wherein
Input picture is converted, fisrt feature figure and the data volume second feature figure big relative to fisrt feature figure are generated
Step includes:
Input picture is filtered, a filtered input picture is generated；
Filtered input picture is converted, generate fisrt feature figure and data volume it is big relative to fisrt feature figure second
Characteristic pattern.
3. image matching method as claimed in claim 2, wherein
Filtered input picture is converted, generate fisrt feature figure and data volume it is big relative to fisrt feature figure second
In the step of characteristic pattern,
Filtered input picture or the second feature figure are converted with different transformation parameters, thus generate from it is different
The corresponding at least two fisrt feature figure of transformation parameter.
4. image matching method as claimed in claim 2, wherein
Based on multiple first related coefficients corresponding with multiple predefined parameters, the value range of the second search parameter is determined
Step includes:
In multiple first related coefficients corresponding with multiple predefined parameters, determining and maximum first related coefficient of value
Corresponding predefined parameter；
Based on predefined parameter corresponding with maximum first related coefficient of value, the value range of the second search parameter is determined.
5. image matching method as described in claim 1, wherein
The first intermediate features figure is generated based on set the first search parameter and fisrt feature figure, calculates generated first
In the step of the first related coefficient between intermediate features figure and the template image,
The pixel in the fisrt feature figure is determined based on the first set search parameter, and merely with identified picture
The value of element generates the first intermediate features figure.
6. a kind of image processing apparatus, comprising:
Converter unit converts input picture, generate fisrt feature figure and data volume it is big relative to fisrt feature figure the
Two characteristic patterns；
First search parameter is respectively set as multiple predefined parameters by the first search unit, based on the first set search ginseng
It counts with fisrt feature figure and generates the first intermediate features figure, calculate between the first intermediate features figure and template image generated
First related coefficient, wherein first related coefficient is corresponding with predefined parameter；
Second search unit is based on multiple first related coefficients corresponding with multiple predefined parameters, determines that the second search is joined
Several value ranges, and second search parameter is set separately in the value range of the second search parameter, based on set
Fixed the second search parameter and second feature figure generates the second intermediate features figure, calculate the second intermediate features figure generated with
The second related coefficient between the template image；
Judging unit, in the case where calculated second related coefficient meets predetermined condition, be determined as the input picture with
The template image matching, based on all second search parameters set in the value range of the second search parameter and second
The second related coefficient between all second intermediate features figures and the template image that characteristic pattern generates all is unsatisfactory for predetermined item
In the case where part, it is determined as that the input picture and the template image mismatch.
7. image processing apparatus as claimed in claim 6, wherein
The converter unit is filtered input picture, generates a filtered input picture, and to filtered defeated
Enter image to be converted, generates fisrt feature figure and the data volume second feature figure big relative to fisrt feature figure.
8. image processing apparatus as claimed in claim 7, wherein
The converter unit converts filtered input picture or the second feature figure with different transformation parameters, from
And generate at least two fisrt feature figure corresponding from different transformation parameters.
9. image processing apparatus as claimed in claim 7, wherein
For second search unit in the first related coefficient corresponding with multiple predefined parameters, determination is maximum with value
The corresponding predefined parameter of first related coefficient, and it is based on predefined parameter corresponding with maximum first related coefficient of value, really
The value range of fixed second search parameter.
10. image processing apparatus as claimed in claim 6, wherein
First search unit determines the pixel in the fisrt feature figure based on the first set search parameter, and
Merely with the value of identified pixel, the first intermediate features figure is generated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510147484.7A CN106156699B (en) | 2015-03-31 | 2015-03-31 | Image processing apparatus and image matching method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510147484.7A CN106156699B (en) | 2015-03-31 | 2015-03-31 | Image processing apparatus and image matching method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106156699A CN106156699A (en) | 2016-11-23 |
| CN106156699B true CN106156699B (en) | 2019-06-25 |
Family
ID=57337240
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510147484.7A Active CN106156699B (en) | 2015-03-31 | 2015-03-31 | Image processing apparatus and image matching method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106156699B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101276411A (en) * | 2008-05-12 | 2008-10-01 | 北京理工大学 | Fingerprint identification method |
| CN101515286A (en) * | 2009-04-03 | 2009-08-26 | 东南大学 | Image matching method based on image feature multi-level filtration |
| CN102087710A (en) * | 2009-12-03 | 2011-06-08 | 索尼公司 | Learning device and method, recognition device and method, and program |
| CN102292745A (en) * | 2009-01-23 | 2011-12-21 | 日本电气株式会社 | image signature extraction device |
| CN103714159A (en) * | 2013-12-27 | 2014-04-09 | 中国人民公安大学 | Coarse-to-fine fingerprint identification method fusing second-level and third-level features |
| CN104268880A (en) * | 2014-09-29 | 2015-01-07 | 沈阳工业大学 | Depth information obtaining method based on combination of features and region matching |
-
2015
- 2015-03-31 CN CN201510147484.7A patent/CN106156699B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101276411A (en) * | 2008-05-12 | 2008-10-01 | 北京理工大学 | Fingerprint identification method |
| CN102292745A (en) * | 2009-01-23 | 2011-12-21 | 日本电气株式会社 | image signature extraction device |
| CN101515286A (en) * | 2009-04-03 | 2009-08-26 | 东南大学 | Image matching method based on image feature multi-level filtration |
| CN102087710A (en) * | 2009-12-03 | 2011-06-08 | 索尼公司 | Learning device and method, recognition device and method, and program |
| CN103714159A (en) * | 2013-12-27 | 2014-04-09 | 中国人民公安大学 | Coarse-to-fine fingerprint identification method fusing second-level and third-level features |
| CN104268880A (en) * | 2014-09-29 | 2015-01-07 | 沈阳工业大学 | Depth information obtaining method based on combination of features and region matching |
Non-Patent Citations (1)
| Title |
|---|
| 快速的多级指纹混合匹配方法;曹国等;《模式识别与人工智能》;20091031;第787-793页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106156699A (en) | 2016-11-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | Robust learning with kernel mean $ p $-power error loss | |
| CN103646244B (en) | Extraction, authentication method and the device of face characteristic | |
| CN106295526B (en) | The method and device of Car image matching | |
| JP2019079553A (en) | System and method for detecting line in vision system | |
| CN105518717B (en) | A kind of face identification method and device | |
| US20140212044A1 (en) | Image Matching Using Subspace-Based Discrete Transform Encoded Local Binary Patterns | |
| CN105718848B (en) | Quality evaluation method and device for fingerprint image | |
| US10832030B2 (en) | Method and apparatus of selecting candidate fingerprint image for fingerprint recognition | |
| US7233700B2 (en) | System and method for signal matching and characterization | |
| CN110348412A (en) | A kind of key independent positioning method, device, electronic equipment and storage medium | |
| Sugimoto et al. | Universal approach for dct-based constant-time gaussian filter with moment preservation | |
| US20200293857A1 (en) | Cnn processing device, cnn processing method, and program | |
| CN108596250A (en) | Characteristics of image coding method, terminal device and computer readable storage medium | |
| Shu et al. | ORB-oriented mismatching feature points elimination | |
| JPH04299469A (en) | Recognizing method for substance in image and application on substance tracking in series of images | |
| CN110188407A (en) | Method and device for determining liquid flow parameters in porous media | |
| CN111222558B (en) | Image processing method and storage medium | |
| CN106156699B (en) | Image processing apparatus and image matching method | |
| Crespo et al. | Revisiting complex moments for 2-D shape representation and image normalization | |
| CN104298980B (en) | A kind of finger print matching method and device based on smart card | |
| Khalil | Reference point detection for camera-based fingerprint image based on wavelet transformation | |
| CN110059708B (en) | Method and device for generating descriptor and storage medium | |
| CN115423855B (en) | Template matching method, device, equipment and medium for image | |
| Martins et al. | Disparity energy model using a trained neuronal population | |
| Cicconet et al. | Complex-valued hough transforms for circles |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |