Method and system for detecting uniformity of distribution of anodized layer

Abstract

The present application discloses a method and system for detecting the uniformity of the distribution of anodized layers, and belongs to the technical field of oxide layer uniformity detection. The method for detecting the uniformity of the distribution of anodized layers includes: Step 1: irradiating the top of the target area with a linear laser, and receiving the reflection spectrum of the linear laser in the target area with a receiver to obtain the first image information; Step 2: obtaining the target image of the area to be detected to obtain the second image information; Step 3: aligning the first image information and the second image information according to the corresponding positions; Step 4: performing a similarity comparison between the first feature extracted from the first image information and the first sample in the first database to obtain a suspected area; wherein the first sample is the feature of the reflection spectrum of the linear laser in the target area of ​​the area with uneven thickness. The technical solution provided by the present application can effectively increase the monitoring efficiency and the monitoring accuracy.

Description

Technical Field

[0001] The present application relates to the technical field of oxide layer uniformity detection, and in particular to a method and system for detecting the distribution uniformity of an anodized layer. Background Art

[0002] In addition to providing aesthetic appeal, metal finishes on electronic products such as laptops and tablets usually require surface reinforcement to protect the housing of these products and increase product life. These metal surfaces are generally decorated and reinforced using anodizing processes.

[0003] For some products with higher requirements, after anodizing, the thickness uniformity of the oxide layer needs to be tested to avoid the appearance of defective products.

[0004] At present, the uniformity of oxide layer thickness is detected by staff relying on empirical rules. However, this detection method is inefficient and not very accurate. Therefore, in some technologies, visual recognition technology is also used to identify and detect the uniformity of oxide layer thickness to replace manual detection.

[0005] Since the characteristics of the areas with inconsistent oxide layer thickness are diverse and do not have obvious characteristics compared with other uniform areas, the accuracy of using visual recognition technology to detect whether the oxide layer is uniform is actually not high. Summary of the invention

[0006] The content of this application is used to introduce concepts in a brief form, which will be described in detail in the detailed implementation section below. The content of this application is not intended to identify the key features or essential features of the technical solution claimed for protection, nor is it intended to limit the scope of the technical solution claimed for protection.

[0007] As a first aspect of the present application, in order to solve the technical problem of low detection accuracy and low detection efficiency of the thickness uniformity of the anodized layer, the present application provides a method for detecting the uniformity of the distribution of the anodized layer, comprising the following steps:

[0008] Step 1: Use a linear laser to irradiate the top of the target area, and use a receiver to receive the reflection spectrum of the linear laser in the target area to obtain first image information;

[0009] Step 2: Obtain a target image of the area to be detected to obtain second image information;

[0010] Step 3: Align the first image information and the second image information with each other according to corresponding positions;

[0011] Step 4: performing a similarity comparison between the first feature extracted from the first image information and the first sample in the first database to obtain a suspected area; wherein the first sample is a feature of the reflected light spectrum of the linear laser in the target area of ​​the uneven thickness area;

[0012] Step 5: extracting a second feature from the suspected area in the second image information, and comparing the second feature with the second sample in the second database for similarity to determine whether the suspected area is an area with uneven thickness;

[0013] The second sample is image information of a region with uneven thickness.

[0014] In the solution provided by the present application, a linear laser is first used to illuminate the target area, and then a reflection spectrum obtained by reflection in the linear laser area is received. The reflection spectrum is first analyzed to obtain the diffuse reflection of the laser in the target area, and to obtain areas in the detection area where there are different laser reflection directions. The areas with different reflection directions in the laser are often areas where there are uneven thickness changes. The visual algorithm is then used to further screen the area to reduce the number of samples and increase the accuracy of sample judgment. In this way, the technical solution provided by the present application can effectively increase detection efficiency and accuracy.

[0015] When the receiver receives reflected light, the reflected light that is incident on the receiver at an angle will cause the area of ​​the light spot on the receiver to become larger. In this way, it is difficult for the receiver to measure the relationship between the light spot received on the receiver and the reflection position on the detection area, thereby reducing accuracy.

[0016] Furthermore, step 1 includes the following steps:

[0017] Step 11: Obtain the width of the area to be detected, and emit a linear laser whose length is longer than the width of the area to be detected;

[0018] Step 12: emit a linear laser to the top of the area to be detected, the linear laser and the surface of the area to be detected are inclined to each other, and a receiver is used to receive the reflected light of the linear laser, wherein the plane of the receiver and the reflected light of the linear laser are perpendicular to each other;

[0019] Step 13: The linear laser is moved downward in a stepwise manner, and the receiver obtains the reflected light of the linear laser in all areas of the detection area to obtain a reflection spectrum, and uses the reflection spectrum as the first image information.

[0020] In the solution provided in the present application, the plane of the receiver and the reflected light of the linear laser are set to be perpendicular to each other. In this way, when receiving the reflected light, it can be ensured that the light irradiated on the smooth part of the surface area can be shot vertically onto the receiver, accurately representing the position of the light reflection area on the detection area, thereby increasing the accuracy of monitoring.

[0021] Further, in step 13, the distance that the linear laser moves each time is the width of the linear laser irradiated on the receiver;

[0022] After acquiring the reflected light, the detector arranges the reflected light from top to bottom in sequence to obtain a reflection spectrum equal to the area of ​​the detection area.

[0023] In the technical solution provided by the present application, the corresponding reflection spectrum is obtained by arranging the reflected light from top to bottom. Therefore, the reflection spectrum is actually the reflection area obtained after each position in the detection area is irradiated with the laser at the same angle. If the light reflected in a certain area is greatly different from the rest of the light, the diffuse reflection of the surface reflecting the light in the detection area must be inconsistent with the surrounding conditions, so there may be uneven thickness.

[0024] Step 4 includes the following steps:

[0025] Step 41: extracting each linear region constituting the first image information based on the Steger centerline extraction algorithm to obtain a first feature of each linear region;

[0026] Step 42: Perform a first sample comparison between the first feature and the first sample in the first database, and take the area where the similarity exceeds a preset value as a suspected area.

[0027] Further, step 41: comprises the following steps:

[0028] Step 411: grayscale processing the first image information;

[0029] Step 412: construct a 2×2 matrix H(x, y) using the second-order derivative of each pixel point (x, y) in each linear region in the first image information:

[0030]

[0031] Among them, I xx ,I xy ,I yy Respectively represent the second-order partial derivatives of the image in different directions at the (x, y) point;

[0032] Calculate the eigenvalues ​​and eigenvectors of the matrix H(x,y), and take the eigenvector with the largest absolute value as the normal direction of the point;

[0033] Calculate the eigenvalues ​​and eigenvectors of the matrix H(x,y), and take the eigenvector with the largest absolute value as the normal direction of the point;

[0034] Step 413: In the normal direction of each pixel point, the grayscale distribution is expanded using Taylor polynomials;

[0035]

[0036] Among them, I(u, v) represents the gray value, (u0, v0) is the initial pixel coordinate, t is the offset along the normal direction, and Represent the gradient and Hessian matrix respectively;

[0037] Among them, I(u, v) represents the gray value, (u0, v0) is the initial pixel coordinate, t is the offset along the normal direction, and Represent the gradient and Hessian matrix respectively;

[0038] Derivate the Taylor expansion and set it to 0, and solve for the value of t, which represents the distance from the initial pixel point along the normal direction to the center line of the light strip;

[0039] Step 414: Output the extracted linear laser centerline image.

[0040] In this solution, features are extracted based on the Steger centerline extraction algorithm.

[0041] By calculating the Hessian matrix of each point in the image, the eigenvalue and eigenvector of the matrix are used to determine the normal direction of the line, and then the extreme point in the normal direction is found as the center of the line. This method based on the second-order derivative can more accurately reflect the changes in details in the image.

[0042] Furthermore, step 5 includes the following steps:

[0043] Step 51: extracting a second feature from the suspected area in the second image information;

[0044] Step 52: Compare the second feature with the second sample in the second database for similarity, and regard the area where the similarity exceeds a preset threshold as an area with uneven thickness.

[0045] Further, step 51 includes the following steps:

[0046] Step 511: Define the suspected region I(x, y) and the scale space L(x, y, σ) as:

[0047] L(x,y,σ)=G(x,y,σ)×I(x,y);

[0048] Where G(x, y, σ) is a Gaussian function;

[0049]

[0050] Step 512: extract feature points from the suspected area, and use the gradient direction histogram within the feature point area to determine one or more main directions, and use the gradient amplitude m and direction θ of each feature point to represent the feature point;

[0051]

[0052] Step 513: construct a feature point set of the suspected area;

[0053] Step 514: Perform similarity matching between the feature point set and the second sample.

[0054] The solution provided in this application uses the SIFT algorithm to extract corresponding feature points, and uses the matching degree of the feature points to realize the similarity calculation between images. It can have the characteristics of scale invariance and high-precision recognition, find the characteristics of uneven areas of the anodized layer, and effectively increase the accuracy of the similarity calculation.

[0055] Furthermore, in step 4 and step 5, the calculation formula for the similarity of feature points is as follows:

[0056] For feature points F1 and F2, the similarity is calculated as follows;

[0057] The similarity is calculated as follows:

[0058] As a second aspect of the present application, some embodiments of the present application provide a detection system, which uses the aforementioned detection method for the uniformity of the distribution of the anodized layer to detect the thickness of the anodized layer;

[0059] The detection system comprises:

[0060] A linear laser, used for emitting a linear laser to an area to be inspected;

[0061] A mobile bracket, used to drive the linear laser to move;

[0062] A receiver, used for receiving reflected light from the area to be detected to obtain a first image area;

[0063] A lens, used for acquiring second image information of the area to be detected;

[0064] The controller is used to realize the relevant calculations of each step in the aforementioned method for detecting the uniformity of the distribution of the anodized layer.

[0065] In the solution provided by the present application, a linear laser is first used to irradiate the target area, and then a reflection spectrum obtained by reflection in the linear laser area is received. By first analyzing the reflection spectrum, the diffuse reflection of the laser in the target area can be obtained, and the areas with different laser reflection directions in the detection area can be obtained. The areas with different reflection directions in the laser are often the areas with uneven thickness changes. For this reason, the visual algorithm is used to further screen the area, which can reduce the types of samples and increase the accuracy of sample judgment. In this way, the technical solution provided by the present application can effectively increase the detection efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] The drawings constituting a part of this application are used to provide a further understanding of this application, so that other features, purposes and advantages of this application become more obvious. The illustrative embodiment drawings and their descriptions of this application are used to explain this application and do not constitute an improper limitation on this application.

[0067] In addition, throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that the components and elements are not necessarily drawn to scale.

[0068] In the attached picture:

[0069] Figure 1 Schematic diagram of a linear laser.

[0070] Figure 2 The present invention is a flow chart of a method for detecting the uniformity of distribution of anodized layers. DETAILED DESCRIPTION

[0071] Embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present application. It should be understood that the drawings and embodiments of the present application are only for exemplary purposes and are not intended to limit the scope of protection of the present application.

[0072] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0073] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments.

[0074] Embodiment 1: A uniform anodized layer thickness detection system includes a linear laser, a mobile bracket, a receiver, a lens, and a controller. Figure 1, the linear laser is used to emit a linear laser. After the laser emitted by the laser is irradiated on the anodized layer, a long strip of bright spot will be formed, and the linear laser is irradiated on the anodized layer at an angle, so the linear laser will be reflected on the receiver. The receiver is a photoelectric receiver, which identifies the intensity of the received light signal and determines whether the reflected light from the laser transmitter is received based on the intensity of the light signal. In practice, in order to reduce the influence of visible light, the laser emitted by the linear laser can be set to a laser outside the visible light frequency band, and the receiver only reads the light signal intensity of the light in a specific frequency band. In this way, the receiver generates a corresponding reflection spectrum based on the strength of the received light signal.

[0075] For example, a linear laser is irradiated on the surface of the anodized layer, and the surface of the anodized layer is regarded as a total reflection surface. Then, a receiver can be arranged on the other side of the layer opposite to the linear laser to receive a long strip of light, and the strength of the light signal is used to render the first image information. Therefore, if the surface of the anodized layer is smooth and has few pits, the shape of the received reflected light spectrum strip is uniform. If the surface of the anodized layer is not smooth, most of the reflection is diffuse reflection, and the spectrum of the reflected light received is messy, and it is impossible to form a uniform spectrum on the receiver. The technical solution of the present application adopts the method of receiving the reflected spectrum of the surface of the anodized layer to determine whether the surface of the anodized layer is smooth.

[0076] The mobile bracket is used to drive the linear laser to move. If the width of the linear laser is too large, the received optical signal will be interfered too much, and the reflection position cannot correspond to the reflection spectrum received on the receiver. Therefore, a mobile bracket is needed to drive the linear laser to move.

[0077] In order to ensure that the reflection spectrum received on the receiver can characterize the reflection characteristics of each position in the area to be tested, the distance that the linear laser moves each time is set to the width of the linear laser on the area to be tested. The linear laser will send a linear laser before each movement, and the corresponding receiver will receive the corresponding reflection spectrum. As the linear laser moves downward until the linear laser moves from the top to the bottom of the area to be tested, multiple reflection spectra will be obtained on the receiver. By splicing all the reflection spectra together, the reflection characteristics of all positions in the area to be tested can be characterized. The lens is mainly used to receive image information, and an industrial camera is selected. The controller is mainly used for related calculation and control work. The above-mentioned devices can be restored by technicians in this field if their functions and uses are known.

[0078] In the solution provided by the present application, a linear laser is first used to irradiate the target area, and then the reflection in the linear laser area is received to obtain a reflection spectrum. The reflection spectrum analysis shows that there are areas with different laser reflection directions in the detection area, and the areas with different reflection directions in the laser are often areas with uneven thickness changes. For this reason, the area is further screened using a visual algorithm, which can reduce the number of samples during screening and increase the accuracy of sample judgment. In this way, the technical solution provided by the present application can effectively increase detection efficiency and accuracy.

[0079] Embodiment 2: The technical solution provided in Embodiment 1 provides a uniform anodized layer thickness detection system, which can detect the thickness of the oxide layer on the surface of the workpiece using the principle of laser reflection. Embodiment 2 provides a method for detecting the uniformity of the distribution of anodized layers on the basis of Embodiment 1, which requires the detection system for the uniformity of the distribution of anodized layers described in Embodiment 1 to detect the thickness of the anodized layer on the surface of the workpiece.

[0080] refer to Figure 2 A method for detecting the uniformity of distribution of anodized layers comprises the following steps:

[0081] Step 1: Arrange the target area to be detected vertically so that the plane of the target area is perpendicular to the horizontal plane, illuminate the top of the target area with a linear laser, and use a receiver to receive the reflection spectrum of the linear laser in the target area to obtain the first image information.

[0082] Furthermore, step 1 includes the following steps:

[0083] Step 11: Fix the workpiece to be inspected so that the plane of the target area on the workpiece to be inspected is perpendicular to the horizontal plane;

[0084] Obtain the width of the area to be detected, and emit a linear laser whose length is longer than the width of the area to be detected;

[0085] Step 12: emit a linear laser to the top of the area to be detected, the linear laser and the surface of the area to be detected are inclined to each other, and a receiver is used to receive the reflected light of the linear laser, wherein the plane of the receiver and the reflected light of the linear laser are perpendicular to each other;

[0086] Step 13: The linear laser is moved downward in a stepwise manner, and the receiver obtains the reflected light of the linear laser in all areas of the detection area to obtain a reflection spectrum, and uses the reflection spectrum as the first image information.

[0087] Furthermore, in step 13, the distance that the linear laser moves each time is the width of the linear laser irradiated on the receiver. After acquiring the reflected light, the detector sequentially arranges the reflected light from top to bottom to obtain a reflection spectrum equal to the area of ​​the detection area.

[0088] In this solution, the first image information obtained is actually the reflection characteristics of the area to be detected. If the area to be detected is smooth, the light signal intensity at the corresponding position in the first image information is uniform. On the contrary, if there is a large pothole area in the area to be detected, the light signal at the corresponding position of the first image information will be missing or weakened due to too much diffuse reflection.

[0089] Step 2: Obtain a target image of the area to be detected to obtain second image information.

[0090] Step 2 mainly uses a camera to take pictures of the area to be inspected. It is sufficient to obtain a standard top view.

[0091] Step 3: Align the first image information and the second image information with each other according to corresponding positions.

[0092] The second image information is directly a top view of the area to be detected, which actually represents the area to be detected. The first image information is composed of reflection areas, so after obtaining the first image information, the reflection area of ​​the first image area needs to be aligned with the second image area.

[0093] For example, point A in the first image information is obtained by irradiating the laser on point a on the area to be detected, so point A in the first image information needs to be aligned with point a in the second image information. Because each point in the area to be detected is irradiated once by the linear laser and needs to reflect the linear laser once, each pixel in the first image information can correspond to each pixel in the second image information.

[0094] Step 4: Compare the first feature extracted from the first image information with the first sample in the first database to obtain a suspected area, wherein the first sample is a feature of the reflection spectrum of the linear laser in the target area of ​​the uneven thickness area.

[0095] In this solution, the samples in the first database are pre-collected samples obtained by irradiating a laser on an area with uneven thickness and receiving reflected light. Therefore, if the feature points in the first image information are similar to the samples in the first database, it indicates that the area where the feature points are located has uneven thickness.

[0096] Step 4 includes the following steps:

[0097] Step 41: extracting each linear region constituting the first image information based on the Steger centerline extraction algorithm to obtain a first feature of each linear region;

[0098] Further, step 41: comprises the following steps:

[0099] Step 411: grayscale processing the first image information;

[0100] Step 412: construct a 2×2 matrix H(x, y) using the second-order derivative of each pixel point (x, y) in each linear region in the first image information:

[0101]

[0102] Among them, I xx ,I xy ,I yy Respectively represent the second-order partial derivatives of the image in different directions at the (x, y) point;

[0103] Calculate the eigenvalues ​​and eigenvectors of the matrix H(x,y), and take the eigenvector with the largest absolute value as the normal direction of the point;

[0104] Step 413: In the normal direction of each pixel point, the grayscale distribution is expanded using Taylor polynomials;

[0105]

[0106] Among them, I(u, v) represents the gray value, (u0, v0) is the initial pixel coordinate, t is the offset along the normal direction, and Represent the gradient and Hessian matrix respectively;

[0107] Derivate the Taylor expansion and set it to 0, and solve for the value of t, which represents the distance from the initial pixel point along the normal direction to the center line of the light strip;

[0108] Step 414: Output the extracted linear laser centerline image.

[0109] Step 42: Perform a first sample comparison between the first feature and the first sample in the first database, and take the area where the similarity exceeds a preset value as a suspected area.

[0110] Step 5: extracting a second feature from the suspected area in the second image information, and comparing the second feature with the second sample in the second database for similarity to determine whether the suspected area is an area with uneven thickness;

[0111] Furthermore, step 5 includes the following steps:

[0112] Step 51: extracting a second feature from the suspected area in the second image information;

[0113] Step 52: Compare the second feature with the second sample in the second database for similarity, and regard the area where the similarity exceeds a preset threshold as an area with uneven thickness.

[0114] Further, step 511: define the suspected region I(x, y), the scale space L(x, y, σ) as:

[0115] L(x,y,σ)=G(x,y,σ)×I(x,y);

[0116] Where G(x, y, σ) is a Gaussian function;

[0117]

[0118] Step 512: extract feature points from the suspected area, and use the gradient direction histogram within the feature point area to determine one or more main directions, and use the gradient amplitude m and direction θ of each feature point to represent the feature point;

[0119]

[0120] The feature points are extracted as follows:

[0121] Each pixel point (sampling point) is compared with all its neighboring points in the image domain (same scale space) and scale domain (adjacent scale space). When the sampling point is larger or smaller than all neighboring points, the point is an extreme point. Usually, each point needs to be compared with 8 neighboring points of the same scale and 9×2 points corresponding to the upper and lower adjacent scales, a total of 26 points.

[0122] Step 513: construct a feature point set of the suspected area;

[0123] Step 514: Perform similarity matching between the feature point set and the second sample.

[0124] The solution provided in this application uses the SIFT algorithm to extract corresponding feature points, and uses the matching degree of the feature points to realize the similarity calculation between images. It can have the characteristics of scale invariance and high-precision recognition, find the characteristics of uneven areas of the anodized layer, and effectively increase the accuracy of the similarity calculation.

[0125] Step 4 and step 5 are actually a process of further determining the existence of areas with uneven thickness. Relatively speaking, step 4 is mainly based on reflection characteristics to discover potential uneven areas. Step 5 is to further determine the areas with uneven thickness. The reason why this scheme is designed in this way is mainly to reduce the number of introduced features. If the scheme of step 5 is used directly for feature matching, many unpredictable features will be introduced in practice. These features are not necessarily areas with uneven thickness, but there will be situations where there is a certain degree of similarity with areas with uneven thickness. In this case, it is necessary to increase the judgment threshold to reduce the impact of such noise areas. If the judgment threshold is increased, the detection accuracy is easily reduced. For this reason, in this scheme, two different schemes are used for screening, which avoids some special light and shadow imitations or situations that affect the judgment of uneven thickness, and increases the accuracy of judgment.

[0126] Furthermore, in both step 4 and step 5, the first sample and the second sample are matched by similarity. The similarity is calculated as cosine similarity, and the calculation formula is as follows:

[0127] The calculation formula of the similarity of feature points is as follows:

[0128] For feature points F1 and F2, the similarity is calculated as follows;

[0129]

[0130] Cosine similarity only considers the direction of the vectors, not their length (modulus). Therefore, even if there are two feature points that need to be compared, even if the corresponding lengths of their vector amplitudes and angles after conversion are different, the similarity calculation can be completed accurately. In this way, in this solution, it can adapt well to the comparison between different first features and first samples. In practice, due to the randomness of diffuse reflection, the vector lengths of the first feature and the first sample are inconsistent. Using other similarity comparison methods will result in low accuracy of similarity calculation, and thus the first feature and the first sample cannot be matched.

[0131] The above description is only some preferred embodiments of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present application is not limited to the technical solutions formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above invention concept. For example, the above features are replaced with (but not limited to) the technical features with similar functions disclosed in the embodiments of the present application to form a technical solution.

Claims

1. A method for detecting the uniformity of distribution of anodized layers, characterized in that: The steps include: Step 1: Arrange the target area to be detected vertically so that the plane of the target area is perpendicular to the horizontal plane, illuminate the top of the target area with a linear laser, and use a receiver to receive the reflection spectrum of the linear laser in the target area to obtain the first image information; Step 2: Obtain a target image of the area to be detected to obtain second image information; Step 3: Align the first image information and the second image information with each other according to corresponding positions; Step 4: performing a similarity comparison between the first feature extracted from the first image information and the first sample in the first database to obtain a suspected area; wherein the first sample is a feature of the reflected light spectrum of the linear laser in the target area of ​​the uneven thickness area; Step 5: extracting a second feature from the suspected area in the second image information, and comparing the second feature with the second sample in the second database for similarity to determine whether the suspected area is an area with uneven thickness; Wherein, the second sample is image information of a region with uneven thickness; Step 1 includes the following steps: Step 11: Fix the workpiece to be inspected so that the plane of the target area on the workpiece to be inspected is perpendicular to the horizontal plane; Obtain the width of the area to be detected, and emit a linear laser with a length greater than the width of the area to be detected; Step 12: emit a linear laser to the top of the area to be detected, the linear laser and the surface of the area to be detected are inclined to each other, and a receiver is used to receive the reflected light of the linear laser, wherein the plane of the receiver and the reflected light of the linear laser are perpendicular to each other; Step 13: The linear laser is moved downward step by step, and the receiver obtains the reflected light of the linear laser in all areas of the detection area to obtain a reflection spectrum, and uses the reflection spectrum as the first image information; Step 5 includes the following steps: Step 51: extracting a second feature from the suspected area in the second image information; Step 52: Compare the second feature with the second sample in the second database for similarity, and regard the area where the similarity exceeds a preset threshold as an area with uneven thickness; Step 51 includes the following steps: Step 511: Define the suspected region I(x, y) and the scale space L(x, y, σ) as: L(x,y,σ)=G(x,y,σ)×I(x,y); Where G(x, y, σ) is a Gaussian function; Step 512: extract feature points from the suspected area, and use the gradient direction histogram in the neighborhood of the feature points to determine one or more main directions, and use the gradient amplitude m and direction θ of each feature point to represent the feature point; Step 513: construct a feature point set of the suspected area; Step 514: performing similarity matching between the feature point set and the second sample; In step 4 and step 5, the calculation formula for the similarity of feature points is as follows: For feature points F1 and F2, the similarity is calculated as follows; The similarity is calculated as follows:

2. The method for detecting the uniformity of the distribution of the anodized layer according to claim 1, characterized in that: In step 13, the distance that the linear laser moves each time is the width of the linear laser irradiated on the receiver; After acquiring the reflected light, the detector arranges the reflected light from top to bottom in sequence to obtain a reflection spectrum equal to the area of ​​the detection area.

3. The method for detecting the distribution uniformity of anodized layers according to claim 1, characterized in that: Step 4 includes the following steps: Step 41: extracting each linear region constituting the first image information based on the Steger centerline extraction algorithm to obtain a first feature of each linear region; Step 42: The first feature is compared with the first sample in the first database, and the area where the similarity exceeds a preset value is regarded as a suspected area.

4. The method for detecting the uniformity of the distribution of the anodized layer according to claim 3, characterized in that: Step 41 includes the following steps: Step 411: grayscale processing the first image information; Step 412: construct a 2×2 matrix H(x, y) using the second-order derivative of each pixel point (x, y) in each linear region in the first image information: Among them, I xx ,I xy ,I yy Respectively represent the second-order partial derivatives of the image in different directions at the (x, y) point; Calculate the eigenvalues ​​and eigenvectors of the matrix H(x,y), and take the eigenvector with the largest absolute value as the normal direction of the point; Step 413: In the normal direction of each pixel, the grayscale distribution is expanded using Taylor polynomials; Among them, I(u, v) represents the gray value, (u0, v0) is the initial pixel coordinate, t is the offset along the normal direction, and Represent the gradient and Hessian matrix respectively; Derivate the Taylor expansion and set it to 0, and solve for the value of t, which represents the distance from the initial pixel point along the normal direction to the center line of the light strip; Step 414: Output the extracted linear laser centerline image.

5. A detection system, characterized in that: The thickness of the anodized layer is detected by using the method for detecting the uniformity of the distribution of the anodized layer as claimed in any one of claims 1 to 4; The detection system comprises: A linear laser, used for emitting a linear laser to an area to be inspected; A mobile bracket, used to drive the linear laser to move; A receiver, used for receiving reflected light from the area to be detected to obtain a first image area; A lens, used for acquiring second image information of the area to be detected; A controller for implementing the various steps in the method for detecting the uniformity of distribution of anodized layers according to any one of claims 1 to 4.

Images