CN116228684A - Method and device for image processing of appearance defect of battery case - Google Patents

Abstract

The invention discloses a battery shell appearance defect image processing method and device, which relate to the technical field of image processing and comprise the following steps: acquiring images of all sides of a target battery shell; synthesizing a plurality of images of the light sources with different angles of the target battery into a plurality of pictures with different types; inputting a standard image template; performing gray scale comparison on a plurality of different types of pictures synthesized by the target battery and a standard image template, and positioning a defect area of the target battery; classifying the defect type of the target battery through deep learning; screening images with defects, and traversing the images with defects into a rule judging algorithm; outputting the defect type, defect size and defect coordinate position of the target shell. The image processing method disclosed by the invention can be suitable for defects of different surface forms and different types, reduces the influence of water stains and greasy dirt on an algorithm, and improves the detection efficiency.

Description

Method and device for image processing of appearance defect of battery case

technical field

The invention relates to the technical field of image processing, in particular to an image processing method and device for appearance defects of a battery casing.

Background technique

Lithium battery shells are mainly divided into three categories: steel shells, aluminum shells, and soft shells. At present, the detection of surface defects on lithium battery shells is still mainly manual inspection. During long-term single repetitive work, workers are prone to fatigue , resulting in unqualified products flowing into the application market. Different workers operate with different degrees of latitude and strictness, and there will be subjective inconsistencies in the judgment of the qualification of extreme parts. The surface shape of the aluminum battery case is complex and the types of defects are diverse, and water stains, oil stains, and processing textures that appear during the production process are easily misidentified as defects, which interferes with the test results.

Contents of the invention

In view of the complex surface shape and various defect types of aluminum battery casings in the prior art, and the water stains, oil stains, and processing textures that appear during the production process are easily misidentified as defects and interfere with the test results, so the present invention provides a A battery case appearance defect image processing method and device, the specific content is as follows:

A method for processing an image of a battery case appearance defect, comprising the following steps:

By acquiring images of each side of the target battery case;

Combining multiple images of light sources from different angles of the target battery into multiple images of different types;

Input standard image template;

Comparing multiple synthesized images of different types with the standard image template in gray scale to locate the defect area of the target battery;

Through deep learning, classify the defect type of the target battery case;

Screen the images with defects, and traverse the rule judgment algorithm for the images with defects;

Output the defect type, defect size and defect coordinate position of the target battery case.

Further, the images of the light sources at different angles are scanned by a 2.5D camera.

Further, the images of the different light sources include a standard map, a first shape map, a second shape map, a gloss ratio map, a diffuse reflection map and a regular reflection map.

Further, the images of each surface of the target battery casing also include images with height information scanned by a 3D camera.

Further, the image with height information includes a 3D height map and a 3D grayscale image.

Further, the combination of the regular reflection image and the diffuse reflection image identifies whether there are scratches or abrasions on the target battery case;

The gloss ratio map, the first shape map and the second shape map are used in combination to identify various defects of the target battery case.

Further, the rule determination algorithm includes the following steps:

Judging whether the image with defects is a small-area defect, and if it is a small-area defect, output the defect type, defect size and defect coordinate position of the target shell;

If the image with defects does not belong to a small-area defect, the defect image is divided into a point type image and a line type image;

The point-type image and the line-type image output defect sizes through corresponding area, length, and number rules.

Further, the point type image calculates the size of the defect through the area rule and the number rule.

Further, the line type image calculates the defect size through the length rule.

Further, the aforementioned image processing method for appearance defects of a battery case includes: a detection system arranged sequentially on the battery case production line; a control component used to control the detection system; wherein, the detection system Including: the first code scanning gun, configured to scan the appearance of the battery case in the length direction; the second code scanning gun, configured to scan the appearance of the battery case in the height direction; the first lateral positioning mechanism, configured to limit the appearance of the battery case The horizontal position on the production line; the Cartesian coordinate mechanical assembly is arranged on the side of the battery case production line; the first visual detection assembly is installed on the rectangular coordinate mechanical assembly, and is set to drive the rectangular coordinate mechanical assembly The lower part moves in three-dimensional space to detect the appearance of the battery case; the laser positioning component is set on the battery case production line for positioning the battery case.

Considering the traditional manual detection method and the traditional visual detection method, the image processing method disclosed by the present invention can adapt to different surface shapes and different types of defects, reduce the influence of water stains and oil stains on the algorithm, and improve the detection efficiency. The same set of algorithm models is applicable to the detection of defects of different types of aluminum die-casting products, realizing one-time labeling training, adapting to all similar defect detection, reducing the operating costs of enterprises, and improving the efficiency of later equipment research and development.

Description of drawings

FIG. 1 is a flow chart of steps in a method for processing an image of a battery casing appearance defect in an embodiment of the present invention;

Fig. 2 is a flow chart of the steps of the rule determination algorithm in the embodiment of the present invention.

Detailed ways

Embodiments of the present disclosure will be described in detail below in conjunction with the accompanying drawings.

Embodiments of the present disclosure are described below through specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Apparently, the described embodiments are only some of the embodiments of the present disclosure, not all of them. The present disclosure can also be implemented or applied through different specific implementation modes, and various modifications or changes can be made to the details in this specification based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that, in the case of no conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in the present disclosure, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present disclosure.

As shown in Figure 1, an image processing method for appearance defects of a battery case includes the following steps:

S1, by acquiring images of each side of the target battery case;

Further, in a preferred embodiment of the present application, the images of the light sources at different angles are scanned by a 2.5D camera.

The 2.5D automatic image measuring instrument uses surface light or contour light to illuminate, passes through the camera through the zoom objective lens, captures the image and then transmits it to the computer screen through the S terminal, and then uses the crosshair generator to generate the video crosshair on the display. The benchmark is to aim at the measured object, and drive the optical ruler through the workbench to move in the XY direction. The multi-function data processor performs data processing, and the software performs calculation to complete the measurement.

Further, in a preferred embodiment of the present application, the images of the different light sources include a standard map, a first shape map, a second shape map, a gloss ratio map, a diffuse reflection map and a regular reflection map.

In the embodiment of this application, a 3D camera is mainly used to detect the flatness of the bottom of the target battery case, the mouth of the large surface, the protrusion of the large surface, and the depression of the large surface, so as to judge the flatness of the large surface; a 2.5D line scan camera is used Detect the appearance of the large surface, bottom surface and side surface of the target battery case.

Further, in a preferred embodiment of the present application, the images of each surface of the target battery case further include an image with height information scanned by a 3D camera.

Further, in a preferred embodiment of the present application, the image with height information includes a 3D height map and a 3D grayscale image.

In the embodiment of this application, the 2.5D camera scans multiple images of light sources from different angles at one time, and in addition, the 3D camera scans and outputs images with height information.

S2. Synthesizing multiple images of light sources at different angles of the target battery into multiple pictures of different types;

Further, in a preferred embodiment of the present application, the combination of the regular reflection image and the diffuse reflection image identifies whether there are scratches or abrasions on the target battery case;

The gloss ratio map, the first shape map and the second shape map are used in combination to identify various defects of the target battery case.

S3, input standard image template;

S4. Comparing the multiple synthesized pictures of different types with the standard image template in gray scale, and locating the defect area of the target battery;

S5. Classify the defect type of the target battery casing through deep learning;

Deep learning is a general term for a class of pattern analysis methods. In terms of specific research content, it mainly involves three types of methods:

(1) A neural network system based on convolution operations, that is, a convolutional neural network (CNN).

(2) Autoencoder neural network based on multi-layer neurons, including Autoencoder and Sparsecoding, which have received widespread attention in recent years.

(3) Pre-training in the form of a multi-layer self-encoding neural network, and further optimizing the deep belief network (DBN) of the weight of the neural network by combining the identification information.

Through multi-layer processing, the initial "low-level" feature representation is gradually transformed into a "high-level" feature representation, and complex classification and other learning tasks can be completed with a "simple model". Therefore, deep learning can be understood as "feature learning" or "representation learning".

In the embodiment of the present application, one type of deep learning method can be applied to classify the defect type of the target battery, and multiple types of deep learning methods can also be used to classify the defect type of the target battery case.

S6. Screening the images with defects, and traversing the rule determination algorithm for the images with defects;

Further, as shown in Figure 2, in a preferred embodiment of the present application, the rule determination algorithm includes the following steps:

S601. Determine whether the image with defects is a small-area defect, and if it is a small-area defect, output the defect type, defect size, and defect coordinate position of the target shell;

S602. If the image with the defect does not belong to a small-area defect, divide the defect image into a dot-type image and a line-type image;

S603, the point type image and the line type image output defect size according to the corresponding area, length, and number rules.

Further, in a preferred embodiment of the present application, the point type image calculates the size of the defect by using an area rule and a number rule.

Further, in a preferred embodiment of the present application, the line type image calculates the size of the defect through a length rule.

S7. Outputting the defect type, defect size and defect coordinate position of the target battery case.

After the defect image is divided into point-type images and line-type images, the preprocessed image X is first divided into several uniform small blocks X={X1, X2,...,Xn}, and each small block can be All are backgrounds, or a combination of backgrounds and defects, but not all defects, then calculate the variance of each small block of images, and arrange them into an ordered sequence {σ21(j)} in order of variance from small to large; then From {σ21(j)}, remove the image block corresponding to the variance value from small to large (the ratio of the area of the small defect to the area of the entire image is less than 0.5), and then the remaining in {σ21(j)} 40% of the variance of , constitutes a new variance sequence. At this time, among the image blocks corresponding to {σ21(j)}, some are image blocks containing defects, and some are image blocks in the background area.

Perform the following operations on each image block corresponding to {σ21(j)}:

First determine the defect target point initially, calculate the mean value of each image block, and record it as fmean, traverse the pixels of each image block from left to right, and from top to bottom. If the pixel fi, j meets the following conditions, it is marked as the background Point: fi, j > fmean.

If the pixel after traversing fi, j satisfies the following conditions: fi, j+1-fi, j<αfmean, it is considered that the pixel fi, j+1 has the same properties as the previous pixel fi, j, and is marked as a background point or Defect target point, otherwise the current pixel is opposite to the previous one.

After the initial traversal, the pixels in each small image are divided into two parts: the background and the defect target. During this process, a small part of the background is misclassified as a defect target, so it is necessary to remove the false targets after the initial traversal. point. The defect target exists in a connected domain of a certain size. In order to judge whether the defect pixel marked in the initial traversal is a real defect, take the current pixel as the core and form a 3×3 window. If at least half of the defects are included pixel, it is considered as a defect point.

Further, in a preferred embodiment of the present application, the above-mentioned image processing method for appearance defects of a battery case includes: a detection system, which is sequentially arranged on the battery case production line; For controlling the detection system; wherein, the detection system includes: a first code scanning gun, configured to scan the appearance of the battery case in the longitudinal direction; a second code scanning gun, configured to scan the appearance of the battery case in the height direction; the first side The orientation positioning mechanism is set to limit the lateral position of the battery case on the production line; the Cartesian coordinate mechanical component is set on the side of the battery case production line; the first visual inspection component is set and installed on the Cartesian coordinate mechanical component, It is set to move in three-dimensional space under the drive of the rectangular coordinate mechanical component, and is used to detect the appearance of the battery case; the laser positioning component is set on the production line of the battery case, and is used to position the battery case.

Those skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiment, which will not be repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or May be integrated into another system, or some features may be ignored, or not implemented. In another point, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, each unit may exist separately physically, or two or more units may be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

If the integrated unit is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or part of the contribution to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , including several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, read-only memory), random access memory (RAM, random access memory), magnetic disk or optical disk, and other media that can store program codes.

Claims (10)

1. A battery case appearance defect image processing method, is characterized in that, comprises the following steps: By acquiring images of each side of the target battery case; Combining multiple images of light sources from different angles of the target battery into multiple images of different types; Input standard image template; Comparing multiple synthesized images of different types with the standard image template in gray scale to locate the defect area of the target battery; Through deep learning, classify the defect type of the target battery case; Screen the images with defects, and traverse the rule judgment algorithm for the images with defects; Output the defect type, defect size and defect coordinate position of the target battery case. 2 . The image processing method for appearance defects of a battery case according to claim 1 , wherein the images of the light sources at different angles are scanned by a 2.5D camera. 3 . 3. The image processing method for appearance defects of a battery case according to claim 1, wherein the images of the different light sources include a standard image, a first shape image, a second shape image, a gloss ratio image, and a diffuse reflection image. diagrams and regular reflection diagrams. 4 . The method for processing an image of an appearance defect of a battery case according to claim 1 , wherein the images of each surface of the target battery case further include an image with height information scanned by a 3D camera. 5 . The method for processing an image of a battery casing appearance defect according to claim 4 , wherein the image with height information includes a 3D height map and a 3D grayscale image. 6 . 6. The method for processing an image of a battery case appearance defect according to claim 3, wherein the regular reflection image and the diffuse reflection image are combined to identify whether there are scratches or abrasions on the target battery case; The gloss ratio map, the first shape map and the second shape map are used in combination to identify various defects of the target battery case. 7. The method for processing an image of a battery case appearance defect according to claim 1, wherein the rule determination algorithm comprises the following steps: Judging whether the image with defects is a small-area defect, and if it is a small-area defect, output the defect type, defect size and defect coordinate position of the target battery case; If the image with defects does not belong to a small-area defect, the defect image is divided into a point type image and a line type image; The point-type image and the line-type image output defect sizes through corresponding area, length, and number rules. 8 . The method for processing an image of a battery casing appearance defect according to claim 7 , wherein the size of the defect is calculated in the dot type image by using an area rule and a number rule. 9 . The method for processing an image of a battery casing appearance defect according to claim 7 , wherein the size of the defect is calculated using a length rule in the line type image. 10 . 10. An image processing device for appearance defects of battery casings, characterized in that it is implemented in the method for processing images of appearance defects of battery casings according to any one of claims 1 to 9, comprising: a detection system, sequentially set On the battery case production line; the control component is used to control the detection system; wherein, the detection system includes: a first code scanning gun, configured to scan the appearance of the battery case in the length direction; a second code scanning gun, configured for To scan the appearance of the battery case in the height direction; the first lateral positioning mechanism is set to limit the lateral position of the battery case on the production line; the Cartesian coordinate mechanical component is set on the side of the battery case production line; the first visual inspection component , set and installed on the rectangular coordinate mechanical assembly, set to move in three-dimensional space under the drive of the rectangular coordinate mechanical assembly, for detecting the appearance of the battery case; the laser positioning assembly is arranged on the battery case production line , for positioning the battery case.

Images