Paper defect detection method and system

Abstract

The invention belongs to the technical field of material surface structure detection, and provides a paper defect detection method and system.The height information of all pixel points on the surface of paper is obtained through an optical distance measuring sensor, then a height distribution diagram is constructed, the fluctuation mode of the surface structure contour of the paper is analyzed in combination with the convolution difference value, the local trend and the change intensity, and the paper defect detection result is obtained. Through curve fitting and trend clustering of structural features, typical surface structure types, such as periodic corrugations and non-periodic wrinkles, can be effectively distinguished, so that automatic identification and structural cause traceability analysis of paper surface microstructures are realized, and meanwhile, accurate identification of abnormal regions is realized. And the universality and the practicability of the system in an actual industrial scene are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of material surface structure detection, and particularly relates to a paper defect detection method and system. BACKGROUND

[0002] In the production process, due to equipment wear, production raw materials or environmental pollution, careless operation and other reasons, some appearance paper defects such as wrinkles, cracks, holes and stains will be caused. In actual production, it is of great significance to monitor the quality of paper production by accurately judging the type of a paper disease in real time and tracing the source. At present, the paper surface defect detection methods mainly include a digital image feature extraction method and a deep learning method. The digital image feature extraction method greatly depends on data, and the data obtained by sampling due to the texture and other problems of paper changes greatly, resulting in poor detection accuracy. The deep learning method based on computer vision has high requirements for data quality, and it is difficult to obtain very ideal data in the actual production environment, also resulting in poor detection accuracy. In order to solve the above problems, a Chinese invention patent named paper defect detection method, device and electronic equipment with the publication number CN114331957B was applied on November 17, 2021, which proposes a paper defect detection method. The patent avoids optical interference in the actual production environment based on the defect detection method of height information, and further excavates the gradient information by height information to expand the information richness and improve the detection accuracy. However, the patent is based on the height value of each pixel point of the paper, that is, the height distribution of the paper surface is used to monitor the micro-profile abnormality of the paper. In the paper defect, there are two defects of paper surface (periodic corrugated structure) corrugation and paper surface (surface structure longitudinal wrinkle) wrinkle, which are both reflected as uneven height distribution on the paper surface. However, the paper surface corrugation defect can be eliminated by re-calendering and heat pressing, while the paper surface wrinkle defect is difficult to improve and most of them need to be recycled. The above method still cannot effectively distinguish the above two defects, affecting the micro-profile of the paper. SUMMARY

[0003] The present application aims to provide a paper defect detection method and system to solve one or more technical problems in the prior art and at least provide a beneficial choice or create conditions.

[0004] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a paper defect detection method is provided, which comprises the following steps:

[0005] S100, obtaining height information corresponding to each pixel point in a to-be-detected paper image;

[0006] S200, screening out abnormal pixel points according to the obtained height information;

[0007] S300, calculating local depth values of all abnormal pixel points;

[0008] S400, classifying the pixel points according to the local depth values;

[0009] S500, judging the defect condition of the paper to be detected according to the classification result.

[0010] Further, in S100, the specific method for obtaining the height information corresponding to each pixel point in the image of the paper to be detected is: placing the ranging sensor horizontally above the paper to be detected, and detecting the distance between the position point of each pixel point of the paper to be detected and the ranging sensor as the height information corresponding to each pixel point of the paper to be detected.

[0011] Further, the ranging sensor is any one of LK-G80 laser scanning sensor, optoNCDT 1420 optical displacement sensor, VK-X1000 laser confocal microscope, ATOS Triple Scan structured light sensor and capaNCDT 6100 capacitive sensor.

[0012] At present, the paper surface defect detection method mainly includes a digital image feature extraction method and a deep learning method, wherein the digital image feature extraction method has a large dependence on data, and the data obtained by sampling the paper due to the texture and the like changes greatly, resulting in poor detection accuracy; the deep learning method based on computer vision has a high requirement on the quality of data, and it is difficult to obtain very ideal data in the actual production environment, also resulting in poor detection accuracy. In order to solve the above problems, a Chinese invention patent named paper defect detection method, device and electronic equipment with the publication number CN114331957B was applied on November 17, 2021, which proposes a paper defect detection method. The patent based on the height information defect detection method avoids optical interference in the actual production environment, and obtains gradient information by further mining the height information, which expands the information richness and improves the detection accuracy. However, the patent is based on the height value of each pixel point of the paper, that is, according to the height distribution of the paper surface to monitor the micro profile abnormality of the paper. In the paper defect, there are two defects of paper surface (periodic corrugated structure) corrugation and paper surface (surface structure longitudinal wrinkle) wrinkle, which are both reflected as uneven height distribution on the paper surface. However, the paper surface corrugation defect can be eliminated by re-calendering and hot pressing, while the paper surface wrinkle defect is difficult to improve and most of them need to be recycled. The above method still cannot effectively distinguish the above two defects, affecting the micro profile of the paper. To solve the above problems, the present application provides the following method: by obtaining the local depth value of each pixel point of the paper to be detected, and classifying it according to the local depth value, the accuracy of identifying the paper defect can be improved.

[0013] Furthermore, in S200, the specific method for filtering out abnormal pixels based on the acquired height information is as follows: traversing the height information corresponding to all pixels and other pixels in the 8-neighborhood of the pixel, forming a height matrix HM with all the height information (if the current pixel is an edge pixel, the height information corresponding to the blank pixels in its 8-neighborhood is supplemented with zeros), with i being the sequence number of the height matrix, where i∈[1,N], and N represents the number of pixels contained in the paper to be detected;

[0014] Traverse all height matrices contained in the paper to be detected, calculate the average value of each element in the height matrix to form a standard matrix, calculate the difference between all height matrices and the standard matrix in turn, use the difference result as the height difference matrix of each pixel point, perform convolution operation on the standard matrix and each height difference matrix in turn, mark the pixel point with the calculation result greater than the average value as an abnormal pixel point, or mark the pixel point with the calculation result less than zero as an abnormal pixel point.

[0015] Furthermore, in S300, the specific method for calculating the local depth values ​​of all abnormal pixels is as follows: traversing all abnormal pixels and their corresponding height matrices in the paper to be detected, adding the height matrices corresponding to the pixels whose convolution operation results are greater than zero to List 1, and updating List 1 according to the size of the operation results, adding the height matrices corresponding to the pixels whose convolution operation results are less than zero to List 2, and updating List 2 according to the size of the operation results;

[0016] Traverse the list List1, sum and average all the matrices in List1, use the average matrix as the current reference matrix A of List1, start from the first height matrix, calculate the matrix difference between the current matrix and the next adjacent matrix, use the difference result as the difference matrix between the current matrix and the next adjacent matrix, perform convolution operation on the difference matrix and the reference matrix, and record the result as the local depth value of the current pixel;

[0017] Traverse the list List2, sum and average all the matrices in List2, use the average matrix as the current reference matrix B of List2, start from the first height matrix, calculate the matrix difference between the current matrix and the next adjacent matrix, use the difference result as the difference matrix between the current matrix and the next adjacent matrix, perform convolution operation on the difference matrix and the reference matrix, and record the result as the local depth value of the current pixel.

[0018] Furthermore, in S400, the specific method of classifying the pixels according to the local depth value is as follows: in List1, the local depth values ​​of all pixels are traversed and the local depth value is recorded as LDV jwherein j represents the serial number of the pixel point in List1, according to the formula Calculate the depth change between adjacent pixel points, PV j represents the depth change between the jth pixel point and the j+1th pixel point, and M represents the number of pixel points contained in List1.

[0019] Let j = 1, and sequentially traverse all PV j according to the formula If I j+1 ≥ I j and I j+1 ≥ I mean , or I j+1 ≤ I j and I j+1 ≤ I mean , the current pixel point is added to the list List2.

[0020] In List2, traverse the local depth values of all pixel points, and record the local depth value as LDV k wherein k represents the serial number of the pixel point in List2, according to the formula Calculate the depth change between adjacent pixel points, PV k represents the depth change between the kth pixel point and the k+1th pixel point.

[0021] Let k = 1, and sequentially traverse all PV k according to the formula If I k+1 < I k and I k+1 < I mean , or I k+1 > I k and I k+1 > I mean , the current pixel point is added to the list List1.

[0022] Mark all pixel points contained in List1 as wrinkle pixel points, mark all pixel points contained in List2 as ripple pixel points, fit the wrinkle pixel points into a wrinkle curve using curve fitting, and fit the ripple pixel points into a ripple curve using curve fitting.

[0023] The beneficial effect of this step is that: the above step extracts surface curve features by convolution and trend modeling of the local height matrix, and classifies and analyzes the surface profile structure based on the geometric shape, fluctuation pattern, and continuous change trend of the fitted curve, thereby determining whether it is a wrinkle structure or a corrugated structure. By introducing a change trend calculation formula based on the local depth value (LDV), the abnormal pixel points in the surface corrugated defect area caused by the longitudinal wrinkle and periodic corrugated structure in the paper surface structure can be effectively identified and classified. PV j is used to measure the depth change amplitude between adjacent pixel points, which not only considers the accumulation of local values, but also captures the absolute deviation of the change slope, thereby effectively depicting the trend of the pixel points moving away from the dense center in space. Subsequently, the change direction and strength of PV j are accumulated to determine whether the current pixel point is in the edge region with significant overall trend change. Through the above method, discrete pixel points far from the local aggregation center can be accurately screened, i.e., feature points with significant local depth value jumps but still belonging to the abnormal category. The distribution of pixel points is determined to be overall uneven, such as whether the trend strength of the front and rear points in the wrinkle area is continuous, and whether the change in the corrugated area is symmetrical. This enhances the defect feature edge recognition capability, especially for defects with fuzzy linear trend but obvious overall diffusion characteristics, such as slight corrugation and edge micro-protrusion. It avoids misclassifying densely distributed normal areas as surface structure abnormal points, improves detection accuracy and robustness, and provides a more accurate data basis for subsequent curve fitting and defect area enclosure through this classification mechanism.

[0024] Further, in S500, the specific method for judging the defect condition of the paper to be detected according to the classification result is: extracting the first pixel point coordinate of the wrinkle curve as A1(X1, Y1), extracting the last pixel point coordinate of the wrinkle curve as A2(X2, Y2), connecting A1 and A2 to form a line segment Line1, wherein the length of the line segment Line1 is |A1A2|, marking the midpoint of Line1 as M1, and drawing a circle C1 with M1 as the center and |A1A2| / 2 as the radius. Detect whether all pixel points on the wrinkle curve are within the circle C1. If not, update the radius to |A1A2| / 2 + |A1M1| - |A2M1|, update the circle C1 with M1 as the center and the updated radius as the radius. Until all pixel points on the wrinkle curve are within the circle C1, extract the area within the circle C1 as the defect area, and upload the image contained in the defect area to the classifier to obtain the defect result.

[0025] The coordinates of the first pixel point of the ripple curve are extracted as B1 (X1, Y1), and the coordinates of the last pixel point of the wrinkle curve are extracted as B2 (X2, Y2). B1 and B2 are connected to form a line segment Line2, where the length of line segment Line2 is |B1B2|, and the midpoint of Line2 is marked as M2. A circle C2 is drawn with M2 as the center and |B1B2| / 2 as the radius. Check whether all pixels on the wrinkle curve are within circle C2. If all pixels on the wrinkle curve are not within circle C2, the radius is updated to |B1B2| / 2+|B1M2|-|B2M2|, and circle C2 is updated with M2 as the center and the updated radius until all pixels on the wrinkle curve are within circle C2. The area within circle C2 is extracted as the defect area, and the image contained in the defect area is uploaded to the classifier to obtain the defect result.

[0026] The beneficial effect of the above steps is: by extracting the coordinates of the first and last pixel points of the wrinkle curve and constructing a circular area based on these points, the possible defect area is preliminarily defined. However, a circle with a single radius may not be able to completely cover all the pixel points on the wrinkle curve. Therefore, it is necessary to continuously update the radius to ensure that all pixel points are included in the circular range, so as to avoid missing any possible defects. The above steps of updating the radius make the detection process more flexible and accurate, and can effectively adapt to curves of different shapes and sizes, thereby avoiding misjudgment or missed detection due to a fixed radius.

[0027] Furthermore, the classifier is a SVM classifier.

[0028] Beneficial effects of the present invention: The present invention not only detects whether there are structural anomalies on the surface of paper, but also further identifies its structural type (such as periodic corrugations, non-periodic wrinkles), thereby reflecting the physical surface morphology formed by the paper during the production process. Specifically, by quantifying the local depth value of each pixel point and combining the depth change trend and difference between it and the adjacent pixels, an accurate pixel classification mechanism is established. Compared with the traditional visual recognition method that relies on grayscale images or color changes for judgment, the microscopic deformation characteristics of the paper surface are directly quantified through height information, avoiding the errors caused by the subjectivity of visual perception. At the same time, the local depth value used is constructed based on the real physical height and is not affected by light, which significantly enhances the stability of the system and can further distinguish its continuity characteristics from spatial distribution differences, thereby overcoming the classification interference caused by high consistency. Even in the case of small paper deformation, shallow longitudinal wrinkles in the surface structure, or irregular periodic corrugation structure, this method can still accurately identify abnormal areas through in-depth analysis of the local height distribution structure, thereby improving the universality and practicality of the system in actual industrial scenarios.

[0029] The application further provides a paper defect detection system, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to run in units of the following system:

[0030] An information acquisition unit is configured to acquire height information corresponding to each pixel point in a paper image to be detected by an optical sensor;

[0031] An anomaly marking unit is configured to screen out abnormal pixel points according to the acquired height information;

[0032] A feature calculation unit is configured to calculate local depth values of all abnormal pixel points;

[0033] A feature classification unit is configured to classify the pixel points according to the local depth values;

[0034] A defect identification unit is configured to determine a defect condition of the paper to be detected according to a classification result.

[0035] The application has the following beneficial effects: The application not only detects whether there is a structural anomaly on the surface of the paper, but also further identifies the structural type (such as periodic corrugation and non-periodic wrinkle), and thus reflects the physical surface morphology formed in the production process of the paper. Specifically, a quantitative calculation is performed on the local depth value of each pixel point, and a depth change trend and difference between the pixel point and adjacent pixel points are combined to establish an accurate pixel point classification mechanism. Compared with a traditional visual recognition method that relies on a grayscale image or color change for judgment, the height information is directly used to quantitatively calculate the microscopic deformation characteristics of the surface of the paper, so that errors caused by subjective visual perception are avoided, the local depth value is constructed based on the real physical height and is not affected by light, the system stability is significantly enhanced, and the continuity characteristics and spatial distribution difference can be further distinguished, so that classification interference caused by the consistency of the height is overcome. Even in the case that the paper is slightly deformed, the longitudinal wrinkle on the surface structure is shallow, or the period of the periodic corrugation structure is irregular, the method can still accurately identify the abnormal area through in-depth analysis of the local height distribution structure, and the universality and practicality of the system in an actual industrial scene are improved. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 Fig. 1 shows a flowchart of a paper defect detection method;

[0037] Figure 2 Fig. 2 shows a structural diagram of a paper defect detection system. DETAILED DESCRIPTION

[0038] Embodiments of the present application are described below in detail with reference to the accompanying drawings, wherein the same or similar components or components having the same or similar functions are denoted by the same or similar reference numerals throughout the drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present application, and cannot be understood as limiting the present application.

[0039] Embodiment 1:

[0040] Figure 1 A flowchart of a paper defect detection method is shown.

[0041] With reference to Figure 1 The present application provides a paper defect detection method, which comprises the following steps:

[0042] S100, obtaining height information corresponding to each pixel point in a paper image to be detected;

[0043] S200, screening out abnormal pixel points according to the obtained height information;

[0044] S300, calculating local depth values of all abnormal pixel points;

[0045] S400, classifying pixel points according to the local depth values;

[0046] S500, judging the defect condition of the paper to be detected according to the classification result.

[0047] Further, in S100, the specific method of obtaining height information corresponding to each pixel point in a paper image to be detected is as follows: a ranging sensor is placed horizontally above the paper to be detected, and the distance between the position point of each pixel point of the paper to be detected and the ranging sensor is detected as the height information corresponding to each pixel point of the paper to be detected.

[0048] Further, the ranging sensor is an LK-G80 laser scanning sensor.

[0049] Further, in S200, the specific method of screening out abnormal pixel points according to the obtained height information is as follows: all pixel points and the height information corresponding to other pixel points in the 8-neighborhood of the pixel points are traversed, a height matrix HM is formed by all height information (if the current pixel point is an edge pixel point, the height information corresponding to blank pixel points in the 8-neighborhood of the edge pixel point is supplemented by zero padding), and i is used as the serial number of the height matrix, wherein i∈[1,N], N represents the number of pixel points contained in the paper to be detected.

[0050] The average value of each element in the height matrix is calculated to form a standard matrix, the difference between all height matrices and the standard matrix is calculated, and the difference result is taken as the height difference matrix of each pixel point. The standard matrix and each height difference matrix are sequentially subjected to convolution operation, and the pixel points with the operation result greater than the average value are marked as abnormal pixel points, or the pixel points with the operation result less than zero are marked as abnormal pixel points.

[0051] Further, in S300, the specific method for calculating the local depth value of all abnormal pixel points is that all abnormal pixel points and their corresponding height matrices in the paper to be detected are traversed, the height matrix corresponding to the pixel point with the convolution operation result greater than zero is added to the list List1, and the list List1 is updated according to the size of the operation result, the height matrix corresponding to the pixel point with the convolution operation result less than zero is added to the list List2, and the list List2 is updated according to the size of the operation result;

[0052] The matrices in List1 are summed and averaged, the average matrix is taken as the reference matrix A of List1 at present, the matrix difference between the current matrix and the next adjacent matrix is calculated from the first height matrix, the difference result is taken as the difference matrix between the current matrix and the next adjacent matrix, the difference matrix and the reference matrix are subjected to convolution operation, and the operation result is recorded as the local depth value of the current pixel point;

[0053] The matrices in List2 are summed and averaged, the average matrix is taken as the reference matrix B of List2 at present, the matrix difference between the current matrix and the next adjacent matrix is calculated from the first height matrix, the difference result is taken as the difference matrix between the current matrix and the next adjacent matrix, the difference matrix and the reference matrix are subjected to convolution operation, and the operation result is recorded as the local depth value of the current pixel point.

[0054] Further, in S400, the specific method for classifying the pixel points according to the local depth value is that in List1, the local depth values of all pixel points are traversed, the local depth value is recorded as LDV j , where j represents the serial number of the pixel point in List1, and the formula The depth change between adjacent pixel points is calculated, PV j , which represents the depth change between the jth pixel point and the j+1th pixel point, and M represents the number of pixel points contained in List1.

[0055] Let j=1, and all PV j are sequentially traversed within the value range of j, and the formula If I j+1 ≥I j And I j+1 ≥I mean , or I j+1 ≤I j And I j+1 ≤I mean , then add the current pixel point to the list List2;

[0056] In List2, traverse the local depth values ​​of all pixels and record the local depth value as LDV k , where k represents the sequence number of the pixel in List2, according to the formula Calculate the depth change between adjacent pixels, PV k Indicates the depth change between the kth pixel and the k+1th pixel;

[0057] Let k = 1, and traverse all PVs in the range of k. k , according to the formula If I k+1 <I k And I k+1 <I mean , or I k+1 >I k And I k+1 >I mean , then add the current pixel point to the list List1;

[0058] Mark all pixels included in List1 as wrinkle pixels, mark all pixels included in List2 as ripple pixels, fit the wrinkle pixels into a wrinkle curve using curve fitting, and fit the ripple pixels into a ripple curve using curve fitting.

[0059] Further, in S500, the specific method for judging the defect condition of the paper to be detected according to the classification result is as follows: the first pixel point coordinate of the wrinkle curve is extracted as A1(X1, Y1), the last pixel point coordinate of the wrinkle curve is extracted as A2(X2, Y2), a line segment Line1 is formed by connecting A1 and A2, wherein the length of the line segment Line1 is |A1A2|, the midpoint of the line segment Line1 is marked as M1, a circle C1 is drawn with M1 as the center and |A1A2| / 2 as the radius, whether all the pixel points on the wrinkle curve are in the circle C1 is detected, if not, the radius is updated as |A1A2| / 2+|A1M1|-|A2M1|, the circle C1 is updated with M1 as the center and the updated radius, until all the pixel points on the wrinkle curve are in the circle C1, the region in the circle C1 is extracted as the defect region, and the image contained in the defect region is uploaded to the classifier to obtain the defect result.

[0060] The first pixel point coordinate of the wrinkle curve is extracted as B1(X1, Y1), the last pixel point coordinate of the wrinkle curve is extracted as B2(X2, Y2), a line segment Line2 is formed by connecting B1 and B2, wherein the length of the line segment Line2 is |B1B2|, the midpoint of the line segment Line2 is marked as M2, a circle C2 is drawn with M2 as the center and |B1B2| / 2 as the radius, whether all the pixel points on the wrinkle curve are in the circle C2 is detected, if not, the radius is updated as |B1B2| / 2+|B1M2|-|B2M2|, the circle C2 is updated with M2 as the center and the updated radius, until all the pixel points on the wrinkle curve are in the circle C2, the region in the circle C2 is extracted as the defect region, and the image contained in the defect region is uploaded to the classifier to obtain the defect result.

[0061] Further, the classifier is an SVM classifier.

[0062] The beneficial effects of the present application are: the present application not only detects whether there is a structure anomaly on the surface of the paper, but also further identifies the structure type (such as periodic corrugation, non-periodic fold), and then reflects the physical surface morphology formed in the production process of the paper. Specifically, by quantitatively calculating the local depth value of each pixel point, combining the depth change trend and difference between adjacent pixel points, an accurate pixel point classification mechanism is established. Compared with the traditional visual recognition method relying on gray scale or color change for judgment, the micro deformation characteristics of the paper surface are directly quantified through height information, avoiding the error caused by the subjectivity of visual perception. At the same time, the local depth value adopted is based on the real physical height and is not affected by light, which significantly enhances the system stability and can further distinguish the continuity characteristics and spatial distribution difference, thereby overcoming the classification interference caused by the consistency of the height. Even in the case that the paper deformation is small, the longitudinal fold on the surface structure is shallow, or the period of the periodic corrugation structure is irregular, the present method can still accurately identify the abnormal area through in-depth analysis of the local height distribution structure, and improve the universality and practicality of the system in the actual industrial scene.

[0063] In addition, the present application also provides an embodiment of a paper defect detection system, as shown in Figure 2 The system structure diagram of the present application is shown, and the embodiment of the paper defect detection system comprises a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above embodiment of the paper defect detection system are realized.

[0064] The system comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program and runs in the following system units:

[0065] The information acquisition unit is used to acquire the height information of each pixel point in the image of the paper to be detected through the optical sensor;

[0066] The abnormal marking unit is used to screen out abnormal pixel points according to the acquired height information;

[0067] The feature calculation unit is used to calculate the local depth value of all abnormal pixel points;

[0068] The feature classification unit is used to classify the pixel points according to the local depth value;

[0069] The defect recognition unit is used to judge the defect condition of the paper to be detected according to the classification result.

[0070] The paper defect detection system can run on a desktop computer, a notebook computer, a palm computer, a cloud server, and the like. The paper defect detection system can run on a system including, but not limited to, a processor and a memory. Those skilled in the art can understand that the example is only an example of the paper defect detection system, and does not limit the paper defect detection system, which can include more or fewer components, or combine certain components, or different components, for example, the paper defect detection system can also include an input / output device, a network access device, a bus, and the like.

[0071] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, and the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor and the like. The processor is a control center of the paper defect detection system running system, and is connected to each part of the paper defect detection system running system through various interfaces and lines.

[0072] The memory can be used to store the computer program and / or modules, and the processor can realize various functions of the paper defect detection system by running or executing the computer program and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area. The program storage area can store an operating system, at least one application program required for a function (such as a sound playing function, an image playing function, and the like), and the like; and the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, and the like), and the like. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, for example, a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state memory devices.

[0073] While the description of the application has been presented in considerable detail and particularity with respect to several embodiments thereof, it is not intended to restrict or in any way limit the scope of the application to such detail and particularity but rather it is intended to cover the intended scope of the application in its full breadth and scope. Furthermore, the foregoing description has been presented for the purposes of exemplification of the embodiments of the application as presently contemplated and is not intended to be exhaustive or to limit the application to the precise steps or embodiment disclosed. Modifications and changes therein and other embodiments apparent to those of ordinary skill in the art are intended to be included within the scope of the application.

Claims

1. A method for detecting paper defects, characterized in that: The method comprises the following steps: S100, obtaining height information corresponding to each pixel in the paper image to be detected; S200, filtering out abnormal pixels based on the acquired height information; S300, calculating the local depth values ​​of all abnormal pixels; S400, classifying pixels according to local depth values; S500: Determine the defect of the paper to be inspected according to the classification result.

2. The method for detecting paper defects according to claim 1, wherein: In S100, the method for obtaining the height information corresponding to each pixel point in the image of the paper to be detected is: placing the ranging sensor horizontally above the paper to be detected, and detecting the distance between the position point where each pixel point of the paper to be detected is located and the ranging sensor as the height information corresponding to each pixel point of the paper to be detected.

3. The method for detecting paper defects according to claim 2, wherein: In S200, the method for filtering out abnormal pixels based on the acquired height information is as follows: traversing the height information corresponding to all pixels and other pixels in the 8-neighborhood of the pixel, forming a height matrix HM with all the height information, with i as the sequence number of the height matrix, where i∈[1,N], N represents the number of pixels contained in the paper to be detected; Traverse all height matrices contained in the paper to be detected, calculate the average value of each element in the height matrix to form a standard matrix, calculate the difference between all height matrices and the standard matrix in turn, use the difference result as the height difference matrix of each pixel point, perform convolution operation on the standard matrix and each height difference matrix in turn, mark the pixel point with the calculation result greater than the average value as an abnormal pixel point, or mark the pixel point with the calculation result less than zero as an abnormal pixel point.

4. The method for detecting paper defects according to claim 3, wherein: In S300, the method for calculating the local depth values ​​of all abnormal pixels is as follows: all abnormal pixels and their corresponding height matrices in the paper to be detected are traversed, the height matrices corresponding to the pixels whose convolution operation results are greater than zero are added to List 1, and List 1 is updated according to the size of the operation results, and the height matrices corresponding to the pixels whose convolution operation results are less than zero are added to List 2, and List 2 is updated according to the size of the operation results; Traverse the list List1, sum and average all the matrices in List1, use the average matrix as the current reference matrix A of List1, start from the first height matrix, calculate the matrix difference between the current matrix and the next adjacent matrix, use the difference result as the difference matrix between the current matrix and the next adjacent matrix, perform convolution operation on the difference matrix and the reference matrix, and record the result as the local depth value of the current pixel; Traverse the list List2, sum and average all the matrices in List2, use the average matrix as the current reference matrix B of List2, start from the first height matrix, calculate the matrix difference between the current matrix and the next adjacent matrix, use the difference result as the difference matrix between the current matrix and the next adjacent matrix, perform convolution operation on the difference matrix and the reference matrix, and record the result as the local depth value of the current pixel.

5. The method for detecting paper defects according to claim 4, characterized in that: In S400, the method for classifying pixels according to local depth values ​​is as follows: in List1, the local depth values ​​of all pixels are traversed and the local depth values ​​are recorded as LDV j , where j represents the sequence number of the pixel in List1, and the depth change between adjacent pixels is calculated as PV according to the formula j , PV j Indicates the depth change between the jth pixel and the j+1th pixel; Let j = 1, and traverse all PVs in the range of j. j , according to the formula If I j+1 ≥I j And I j+1 ≥I mean , or I j+1 ≤I j And I j+1 ≤I mean , then add the current pixel point to List2, where M represents the number of pixels contained in List1; In List2, traverse the local depth values ​​of all pixels and record the local depth value as LDV k , where k represents the sequence number of the pixel in List2, according to the formula Calculate the depth change between adjacent pixels, PV k Indicates the depth change between the kth pixel and the k+1th pixel; Let k = 1, and traverse all PVs in the range of k. k , according to the formula If I k+1 <I k And I k+1 <I mean , or I k+1 >I k And I k+1 >I mean , then add the current pixel point to the list List1; Mark all pixels included in List1 as wrinkle pixels, mark all pixels included in List2 as ripple pixels, fit the wrinkle pixels into a wrinkle curve using curve fitting, and fit the ripple pixels into a ripple curve using curve fitting.

6. The method for detecting paper defects according to claim 4, characterized in that: In S500, the method for determining the defect of the paper to be detected based on the classification result is as follows: extract the coordinates of the first pixel point of the wrinkle curve as A1 (X1, Y1), extract the coordinates of the last pixel point of the wrinkle curve as A2 (X2, Y2), connect A1 and A2 to form a line segment Line1, where the length of the line segment Line1 is |A1A2|, mark the midpoint of Line1 as M1, draw a circle C1 with M1 as the center and |A1A2| / 2 as the radius, and detect whether all pixel points on the wrinkle curve are within the circle C1. If all pixel points on the wrinkle curve are not within the circle C1, then update the radius to |A1A2| / 2+|A1M1|-|A2M1|, and update the circle C1 with M1 as the center and the updated radius until all pixel points on the wrinkle curve are within the circle C1. Extract the area within the circle C1 as the defect area, and upload the image contained in the defect area to the classifier to obtain the defect result. The coordinates of the first pixel point of the ripple curve are extracted as B1 (X1, Y1), and the coordinates of the last pixel point of the wrinkle curve are extracted as B2 (X2, Y2). B1 and B2 are connected to form a line segment Line2, where the length of line segment Line2 is |B1B2|, and the midpoint of Line2 is marked as M2. A circle C2 is made with M2 as the center and |B1B2| / 2 as the radius. It is detected whether all the pixel points on the wrinkle curve are in circle C2. If all the pixel points on the wrinkle curve are not in circle C2, the radius is updated to |B1B2| / 2+|B1M2|-|B2M2|, and circle C2 is updated with M2 as the center and the updated radius until all the pixel points on the wrinkle curve are in circle C2. The area inside circle C2 is extracted as the defect area, and the image contained in the defect area is uploaded to the classifier to obtain the defect result.

7. A paper defect detection system, characterized in that: The paper defect detection system includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the paper defect detection method according to any one of claims 1 to 6 are implemented.

Images