PCB welding quality detection method and time control method based on infrared imaging

Abstract

The invention relates to the field of wave soldering, in particular to a PCB welding quality detection method and time control method based on infrared imaging, and the method comprises the steps: obtaining an infrared image of a non-welding surface of a PCB; processing and analyzing the infrared image to obtain a PCB non-welding surface space temperature distribution signal and a pin edge gray scale gradient function; a feedback signal is generated according to the temperature signal of the non-welding surface of the PCB and transmitted to the central controller to adjust the transmission speed of the transmission guide rail driven by the motor to transmit the PCB; and identifying a potential defect area existing in the PCB pin welding spot according to the pin edge gray scale gradient function. According to the method, control over the tin immersion time of wave soldering and defect identification are both considered, and the soldering quality of PCB wave soldering is improved.

Description

Technical Field

[0001] This invention relates to the field of wave soldering, and more specifically, to a method for inspecting the soldering quality of PCB boards and a time control method based on infrared imaging. Background Technology

[0002] The soldering time during wave soldering directly affects the soldering quality of PCB boards. Currently, the common method for measuring immersion time in wave soldering is the oven temperature profile test. This method requires preparing a dedicated temperature measuring board with the same PCB material and thickness as the PCB to be produced. A certain number (usually no less than three) of thermocouples are placed on the motherboard, positioned at locations with high density of small components (such as resistors and capacitors), locations of large heat capacity components (such as power modules, ground planes, etc.), and locations of critical components and closely spaced components (such as QFN heat sinks and BGA central solder balls). These thermocouples are fixed to the board surface using high-temperature tape or red glue. An oven temperature measuring instrument is connected to the thermocouples and the wave soldering is performed. After completion, the oven temperature measuring instrument is connected to a computer, and the oven temperature profile parameters for preheating and soldering are read using specialized software and displayed on the interface. This method involves a complex temperature measurement process and cannot perform temperature testing on every PCB board, nor can it inspect the quality of each board.

[0003] Existing technology discloses a wave solder joint defect detection method and system, which solves the problem of unsatisfactory detection accuracy of current wave solder joint defect detection methods based on automatic optical inspection. First, PCBA board template images are acquired, preprocessed, and subjected to color image thresholding. Then, based on the R, G, B color components of the RGB color model and the H, S, V color components of the HSV color model, color image thresholding is performed on the detection area simultaneously, optimizing the thresholding process to facilitate effective extraction of defect areas. A circular template matching optimization method is designed to optimize the positioning of solder and pads in wave solder joints, providing more accurate defect areas for the subsequent second thresholding. Finally, the feature parameters of the acquired solder joint image are compared with the feature parameters of the original image to detect defective samples. However, this method cannot simultaneously control wave soldering time and detect quality. Summary of the Invention

[0004] The purpose of this invention is to disclose a PCB board soldering quality inspection method and time control method based on infrared imaging that takes into account both wave soldering time control and quality inspection.

[0005] To achieve the above objectives, the present invention provides a wave soldering quality inspection method based on infrared imaging, comprising:

[0006] Obtain an infrared image of the non-soldered side of the PCB board;

[0007] The infrared image is processed and analyzed to obtain the gray-scale gradient function of the pin edge;

[0008] Potential defect areas in PCB board pin solder joints are identified based on the gray-scale gradient function of the pin edge.

[0009] Furthermore, the method for obtaining an infrared image of the non-soldering side of the PCB board is as follows: an infrared thermal imager is installed directly above the guide rail that carries the PCB board as it passes through the double-peak solder flow. When the PCB board passes through the double-peak solder flow, the infrared thermal imager captures an image of the non-soldering side of the PCB board to obtain an infrared image of the non-soldering side of the PCB board.

[0010] Furthermore, the method for obtaining the grayscale gradient function at the pin edge is as follows:

[0011] The infrared images are binarized and edge point sequences are extracted based on image edge detection.

[0012] Based on the acquired infrared images, perform infrared image grayscale gradient calculation, thermal gradient parameter extraction, and gradient function extraction;

[0013] The gray-scale gradient function of the pin edge is obtained by calculating the gray-scale gradient of the edge point sequence and the gray-scale value gradient of the infrared image, extracting the thermal gradient parameter, and extracting the gradient function.

[0014] Furthermore, the method for binarizing the acquired infrared images and extracting edge point sequences based on image edge detection is as follows: an image preprocessing algorithm is used to filter out noise from the infrared images to obtain the filtered image;

[0015] The filtered image is binarized and edge detection algorithms are used to extract edge points;

[0016] Arrange all extracted edge points clockwise to obtain the edge sequence P(i), where i = 1, 2, ..., n. P(n) is the endpoint P(x). n ,y n ).

[0017] Furthermore, the method for calculating the grayscale gradient of an infrared image is as follows:

[0018] In Euclidean space, the calculation is as follows:

[0019]

[0020] In the formula, gradf(x,y) represents the gray-level gradient of the image, and |gradf(x,y)| represents the norm of the image gradient. Each pixel in a digital image has derivatives in eight directions. The gradient norms of the derivatives in the x and y directions are combined to obtain the gray-level gradient of the infrared image. Since a digital image is a two-dimensional discrete function f(x,y), the first-order differential is used instead of the first-order derivative. The specific calculation is as follows:

[0021] z = f(x,y), x∈(0,T)y∈(0,T)

[0022]

[0023] h(x,y)=|f(x,y)-m(x,y)|+|f(x,y)-n(x,y)|

[0024] In the formula, the original image size is z(T×T), and m(x) R ,y R ) and n(x L ,y L ) represent the difference images excluding the first column and first row of the original image z, respectively, m(512,y) R ) and n(x R ,512) represent the last column and last row of the difference image, respectively; matrix m(x R ,y R ),n(x L ,y L ),m(T,x L ),n(x L The difference matrices m(x,y) and n(x,y) of the infrared image are formed by merging the first-order differentials into (T,T). The first-order differentials are used to replace the derivatives, and the gradient norms of the differentials of x and y are calculated. The gray-level gradient of the infrared image is calculated through h(x,y), and the corresponding non-negativity constraints are applied to the difference variables to obtain the correct gray-level gradient norm.

[0025] Furthermore, the calculation of the grayscale gradient of the infrared image also includes: optimization, that is, squaring the components of the difference along the X and Y directions, as follows:

[0026] G(x,y)=[f(x,y)-m(x,y)] 2 +[f(x,y)-n(x,y)] 2 =|gradf(x,y)| 2

[0027] Furthermore, based on the edge point sequence and the calculation of the gray-level gradient of the infrared image, the extraction of thermal gradient parameters, and the extraction of the gradient function, the gray-level gradient function of the pin edge is obtained, and its expression is as follows:

[0028] G(x i ,y i )=|gradP(i)| 2 =|gradf(x) i ,y i )| 2

[0029] Furthermore, identifying potential defect areas in the pin soldering joints of the PCB board based on the pin edge gray gradient function includes:

[0030] Taking the boundary point sequence as the abscissa and its corresponding gray gradient as the ordinate, a boundary gradient function is established. There are local minima and local maxima in the entire boundary gradient function. When this local minimum is less than a certain threshold V min or this local maximum is greater than a certain threshold V max defects will appear in the areas where the local minimum and local maximum are located. The thresholds V min and V max need to be obtained through specific experiments; since the fluctuations in the gradient function will increase the difficulty of extracting some characteristic parameters, polynomial fitting is used to smooth the original gradient function. The specific calculation is as follows:

[0031] T(x) = t1x n +t2x n-1 +t3x n-2 +,...,+t n-1 x 2 +t n x 1 .

[0032] For the pin soldering joints: The maximum inter-class variance method is used to segment the original image. When the pixel f(x,y) in the image ≥ T, this point is determined to be a pin soldering joint; when a certain pixel point f(x,y) < T, let f(x,y) = 0, and this point appears as black in the image and is classified as a non-pin soldering joint; T is the set threshold, and f(x,y) is the original image.

[0033] The specific calculation of obtaining the segmentation threshold T of the original image using the maximum inter-class variance method is as follows:

[0034] Assume that the obtained original image has L gray levels, and the number of pixel levels of gray level i is n i , MN represents the total number of pixels. For any pixel point g(x,y), the probability p that the gray level of this point is i is:

[0035]

[0036] Assume that the image threshold T(k) = k taken for image segmentation, where 0 < k < L - 1. This threshold divides the image into two parts, C1 and C2. C1 represents the background, and its gray level is within the range of [0,k]; C2 represents the target to be extracted, and the gray level is within the range of [k + 1, L - 1]. For any pixel point, the probability P A that this pixel point belongs to the background part is:

[0037]

[0038] Randomly select a pixel point, and the probability P that this pixel point belongs to the extraction target B is as follows:

[0039]

[0040] Using the above formula, the average gray values ω of parts C1 and C2 can be obtained A and ω B are as follows:

[0041]

[0042]

[0043] where ω0 is the global gray mean value. After obtaining ω A , ω<www. B , ω0, the between-class variance σ 2 can be calculated:

[0044] σ 2 = p A (ω A ) 2 + p B (ω B - ω0) 2

[0045] After calculating the segmentation threshold, perform corresponding threshold segmentation on the infrared image. The specific calculation is as follows:

[0046]

[0047] When the pixel f(x, y) in the image ≥ T, determine that this point is a pin solder joint; when the pixel f(x, y) in the image < T, determine that this point is a non-pin solder joint

[0048] Calculate the number S of pixel points in each target area. For the corresponding pin solder joint area, obtain the range of the number of pixels in the high-quality solder joint area through experiments (S min , S max ). When S < S min , appropriately slow down the speed of the conveyor rail and increase the tin dipping time of the PCB board. When S > S max , appropriately speed up the speed of the conveyor rail and reduce the tin dipping time of the PCB board

[0049] The present invention provides a method for controlling the soldering time of a PCB board based on infrared imaging, including:<00www.

[0050] Obtain an infrared image of the non-soldering side of the PCB board

[0051] Infrared images are processed and analyzed to obtain the spatial temperature distribution signal of the non-soldering surface of the PCB board;

[0052] A feedback signal is generated based on the temperature signal of the non-soldering surface of the PCB board. This feedback signal is then transmitted to the computer, which adjusts the conveyor rails to control the motor speed, thereby controlling the soldering time of the wave soldering process.

[0053] Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

[0054] This invention performs real-time imaging of PCB wave soldering, then processes and analyzes the infrared images to obtain the temperature signal of the non-soldering surface of the PCB and the gray-scale gradient function of the pin edges. Based on the temperature signal of the non-soldering surface, a feedback signal is generated and transmitted to the central controller to adjust the conveying speed of the guide rail that drives the motor to transport the PCB. The gray-scale gradient function of the pin edges is used to identify potential defect areas in the solder joints of the PCB pins. This achieves both control of the immersion time in wave soldering and defect identification, improving the soldering quality of PCB wave soldering. Attached Figure Description

[0055] Figure 1 This is a flowchart of the PCB board welding quality inspection method based on infrared imaging as described in Example 1;

[0056] Figure 2 This is a diagram of the wave soldering temperature control system based on infrared imaging described in Embodiment 2;

[0057] Figure 3 This is a flowchart of the PCB board welding quality inspection method based on infrared imaging as described in Example 3;

[0058] Figure 4 This is a flowchart of the wave soldering time control and pin solder joint edge gradient function extraction based on infrared images as described in Example 3; Detailed Implementation

[0059] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this patent.

[0060] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0061] Example 1:

[0062] This embodiment provides, as follows: Figure 1 The infrared imaging-based wave soldering quality inspection method shown includes:

[0063] Obtain an infrared image of the non-soldered side of the PCB board;

[0064] The infrared image is processed and analyzed to obtain the gray-scale gradient function of the pin edge;

[0065] Potential defect areas in PCB board pin solder joints are identified based on the gray-scale gradient function of the pin edge.

[0066] This embodiment performs real-time imaging of the PCB wave soldering process, then processes and analyzes the infrared images to obtain the temperature signal of the non-soldering surface of the PCB and the grayscale gradient function of the pin edges. Based on the temperature signal of the non-soldering surface, a feedback signal is generated and transmitted to the central controller to adjust the conveying speed of the guide rail that drives the PCB. The grayscale gradient function of the pin edges is used to identify potential defect areas in the PCB pin solder joints. This achieves both control of the immersion time in wave soldering and defect identification, improving the soldering quality of the PCB wave soldering process.

[0067] Example 2:

[0068] This embodiment further discloses information based on Embodiment 1:

[0069] Furthermore, the method for obtaining an infrared image of the non-soldering side of the PCB board is as follows: an infrared thermal imager is installed directly above the guide rail that carries the PCB board as it passes through the double-peak solder flow. When the PCB board passes through the double-peak solder flow, the infrared thermal imager captures an image of the non-soldering side of the PCB board to obtain an infrared image of the non-soldering side of the PCB board.

[0070] Furthermore, the method for obtaining the grayscale gradient function at the pin edge is as follows:

[0071] The infrared images are binarized and edge point sequences are extracted based on image edge detection.

[0072] Based on the acquired infrared images, perform infrared image grayscale gradient calculation, thermal gradient parameter extraction, and gradient function extraction;

[0073] The gray-scale gradient function of the pin edge is obtained by calculating the gray-scale gradient of the edge point sequence and the gray-scale value gradient of the infrared image, extracting the thermal gradient parameter, and extracting the gradient function.

[0074] Furthermore, the method for binarizing the acquired infrared images and extracting edge point sequences based on image edge detection is as follows: an image preprocessing algorithm is used to filter out noise from the infrared images to obtain the filtered image;

[0075] The filtered image is binarized and edge detection algorithms are used to extract edge points;

[0076] Arrange all extracted edge points clockwise to obtain the edge sequence P(i), where i = 1, 2, ..., n. P(n) is the endpoint P(x). n ,y n ).

[0077] Furthermore, the method for calculating the grayscale gradient of an infrared image is as follows:

[0078] In Euclidean space, the calculation is as follows:

[0079]

[0080] In the formula, gradf(x,y) represents the gray-level gradient of the image, and |gradf(x,y)| represents the norm of the image gradient. Each pixel in a digital image has derivatives in eight directions. The gradient norms of the derivatives in the x and y directions are combined to obtain the gray-level gradient of the infrared image. Since a digital image is a two-dimensional discrete function f(x,y), the first-order differential is used instead of the first-order derivative. The specific calculation is as follows:

[0081] z = f(x,y), x∈(0,T)y∈(0,T)

[0082]

[0083] h(x,y)=|f(x,y)-m(x,y)|+|f(x,y)-n(x,y)|

[0084] In the formula, the original image size is z(T×T), and m(x) R ,y R ) and n(x L ,y L ) represent the difference images excluding the first column and first row of the original image z, respectively, m(512,y) R ) and n(x R ,512) represent the last column and last row of the difference image, respectively; matrix m(x R ,y R ),n(x L ,y L ),m(T,x L ),n(x L The difference matrices m(x,y) and n(x,y) of the infrared image are formed by merging the first-order differentials into (T,T). The first-order differentials are used to replace the derivatives, and the gradient norms of the differentials of x and y are calculated. The gray-level gradient of the infrared image is calculated through h(x,y), and the corresponding non-negativity constraints are applied to the difference variables to obtain the correct gray-level gradient norm.

[0085] Furthermore, the calculation of the grayscale gradient of the infrared image also includes: optimization, that is, squaring the components of the difference along the X and Y directions, as follows:

[0086] G(x,y)=[f(x,y)-m(x,y)] 2 +[f(x,y)-n(x,y)] 2 =|gradf(x,y)|2

[0087] Further, according to the edge point sequence, infrared image gray value gradient calculation, thermal gradient parameter extraction, and extraction gradient function, a pin edge gray gradient function is obtained, and its expression is as follows:

[0088] G(x i ,y i ) = |gradP(i)| 2 = |gradf(x i ,y i )| 2

[0089] Further, identifying potential defect areas existing in the pin solder joints of the PCB board according to the pin edge gray gradient function includes:

[0090] Taking the boundary point sequence as the abscissa and its corresponding gray gradient as the ordinate, a boundary gradient function is established. There are local minima and local maxima in the entire boundary gradient function. When this local minimum is less than a certain threshold V min or this local maximum is greater than a certain threshold V max , defects will appear in the areas where the local minimum and local maximum are located. The thresholds V min and V max need to be obtained through specific experiments; since the fluctuations of the gradient function will increase the difficulty of extracting some characteristic parameters, polynomial fitting is used to smooth the original gradient function, and the specific calculation is as follows:

[0091] T(x) = t1x n +t2x n-1 +t3x n-2 +,...,+t n-1 x 2 +t n x 1 .

[0092] The pin solder joint is: Using the maximum inter-class variance method to perform image segmentation on the original image. When the pixel f(x,y) in the image ≥ T, it is determined that this point is a pin solder joint. When a certain pixel point f(x,y) < T, let f(x,y) = 0, and this point appears black in the image and is classified as a non-pin solder joint; T is the set threshold, and f(x,y) is the original image.

[0093] The specific calculation of using the maximum inter-class variance method to find the segmentation threshold T of the original image is as follows:

[0094] Assume that the obtained original image has L gray levels, and the number of pixel levels of gray level i is n i, MN represents the total number of pixels. Arbitrarily select a pixel point g(x, y), and the probability p that the gray level of this point is i is:

[0095]

[0096] Assume that the image threshold T(k) = k taken for image segmentation, where 0 < k < L - 1. This threshold divides the image into two parts, C1 and C2. C1 represents the background, and its gray level is within the range of [0, k]; C2 represents the target to be extracted, and the gray level is within the range of [k + 1, L - 1]. Arbitrarily select a pixel point, and the probability P that this pixel point belongs to the background part A is:

[0097]

[0098] Arbitrarily select a pixel point, and the probability P that this pixel point belongs to the extracted target B is:

[0099]

[0100] Using the above formula, the average gray values ω A and ω B of the C1 and C2 parts can be obtained as:

[0101]

[0102] where ω0 is the global gray mean value. After obtaining ω A , ω B , ω0, the between-class variance σ 2 can be calculated as:

[0103] σ 2 = p A (ω0 - ω A ) 2 + p B (ω B - ω0) 2

[0104] After calculating the segmentation threshold, perform corresponding threshold segmentation on the infrared image. The specific calculation is as follows:

[0105]

[0106] When the pixel f(x, y) in the image ≥ T, determine that this point is a pin solder joint; when the pixel f(x, y) in the image < T, determine that this point is a non-pin solder joint.

[0107] Calculate the number S of pixel points in each target area. For the corresponding pin solder joint area, obtain the range of the number of pixels in the high-quality solder joint area through experiments (S min , Smax ), when S < S min , appropriately slow down the speed of the transfer guide rail and increase the tin dipping time of the PCB. When S > S max , appropriately speed up the speed of the transfer guide rail and reduce the tin dipping time of the PCB.

[0108] The wave soldering temperature control system device based on infrared imaging is as Figure 2 shown

[0109] In this embodiment, real-time imaging of the wave soldering of the PCB is performed, and then the infrared image is processed and analyzed to obtain the temperature signal of the non-welding surface of the PCB and the gray gradient function of the pin edge. A feedback signal is generated according to the temperature signal of the non-welding surface of the PCB and transmitted to the central controller to adjust the transmission speed of the guide rail for driving the PCB by the motor. The potential defect areas existing in the pin soldering points of the PCB are identified according to the gray gradient function of the pin edge. Thus, the control of the tin dipping time of wave soldering and defect identification are兼顾, and the welding quality of the wave soldering of the PCB is improved.

[0110] Embodiment 3:

[0111] The present invention provides a method for controlling the welding time of a PCB based on infrared imaging as Figure 3 shown, including:

[0112] Obtain the infrared image of the non-welding side of the PCB;

[0113] Process and analyze the infrared image to obtain the temperature signal of the pin soldering points on the non-welding surface of the PCB;

[0114] Generate a feedback signal according to the temperature signal of the non-welding surface of the PCB, and the feedback signal is transmitted to the computer. The computer adjusts the rotation speed of the motor controlling the transfer guide rail to achieve the welding time of wave soldering;

[0115] Control the pin soldering points as follows: perform image segmentation on the original image using the maximum inter-class variance method. When the pixel f(x, y) in the image ≥ T, determine that this point is a pin soldering point. When a certain pixel point f(x, y) < T, set f(x, y) = 0, and this point appears black in the image and is classified as a non-pin soldering point; T is the set threshold, and f(x, y) is the original image.

[0116] This embodiment performs real-time imaging of the PCB wave soldering process, then processes and analyzes the infrared images to obtain the temperature signal of the non-soldering surface of the PCB and the grayscale gradient function of the pin edges. Based on the temperature signal of the non-soldering surface, a feedback signal is generated and transmitted to the central controller to adjust the conveying speed of the guide rail that drives the PCB. The grayscale gradient function of the pin edges is used to identify potential defect areas in the PCB pin solder joints. This achieves both control of the immersion time in wave soldering and defect identification, improving the soldering quality of the PCB wave soldering process.

[0117] The flowchart for wave soldering time control and pin solder joint edge gradient function extraction based on infrared images is as follows: Figure 4 As shown

[0118] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A PCB board soldering quality inspection method based on infrared imaging, characterized in that, Including: Obtain an infrared image of the non-welded side of the PCB board; Process and analyze the infrared image to obtain the pin edge gray-scale gradient function; Identify potential defect areas existing in the pin solder joints of the PCB board according to the pin edge gray-scale gradient function.

2. The PCB board soldering quality inspection method based on infrared imaging according to claim 1, characterized in that, The method for obtaining an infrared image of the non-welded side of the PCB board is as follows: Install an infrared thermal imager directly above the double-peak tin flow where the guide rail for transporting the PCB board passes. When the PCB board passes through the double-peak tin flow, the infrared thermal imager takes an image of the non-welded side of the PCB board to obtain an infrared image of the non-welded side of the PCB board.

3. The PCB board soldering quality inspection method based on infrared imaging according to claim 1, characterized in that, The method for obtaining the pin edge gray-scale gradient function is as follows: Perform binary processing on the collected infrared image and detect the image edge to extract the edge point sequence; Calculate the gray-scale value gradient of the infrared image, extract the thermal gradient parameters and extract the gradient function according to the collected infrared image; Obtain the pin edge gray-scale gradient function based on the edge point sequence, the gray-scale value gradient calculation of the infrared image, the thermal gradient parameter extraction and the extraction of the gradient function.

4. The PCB board soldering quality inspection method based on infrared imaging according to claim 2, characterized in that, The method for performing binary processing on the collected infrared image and detecting the image edge to extract the edge point sequence is as follows: Use an image preprocessing algorithm to filter the noise of the infrared image to obtain the filtered image; Perform binary processing on the filtered image and use an edge detection algorithm to extract the edge points; Arrange all extracted edge points clockwise to obtain the edge sequence P(i), where i = 1, 2, ..., n. P(n) is the endpoint P(x). n ,y n ).

5. The PCB board soldering quality inspection method based on infrared imaging according to claim 2, characterized in that, The method for calculating the gray-scale value gradient of the infrared image is as follows: In the Euclidean space, the calculation is as follows: In the formula, gradf(x,y) represents the gray-scale gradient of the image, and |gradf(x,y)| represents the norm of the image gradient; in a digital image, each pixel point has derivatives in eight directions. Combine the gradient norms of the derivatives in the x and y directions as the gray-scale value gradient of the infrared image. Since a digital image is a two-dimensional discrete function f(x,y), the first-order differential is used to replace the first-order derivative. The specific calculation is as follows: z = f(x,y), x ∈ (0,T) y ∈ (0,T) h(x,y) = |f(x,y) - m(x,y)| + |f(x,y) - n(x,y)| In the formula, the original image size is z(T×T), and m(x) R ,y R ) and n(x L ,y L ) represent the difference images excluding the first column and first row of the original image z, respectively, m(512,y) R ) and n(x R ,512) represent the last column and last row of the difference image, respectively; matrix m(x R ,y R ),n(x L ,y L ),m(T,x L ),n(x L The difference matrices m(x,y) and n(x,y) of the infrared image are formed by merging the first-order differentials into (T,T). The first-order differentials are used to replace the derivatives, and the gradient norms of the differentials of x and y are calculated. The gray-level gradient of the infrared image is calculated through h(x,y), and the corresponding non-negativity constraints are applied to the difference variables to obtain the correct gray-level gradient norm.

6. The PCB board soldering quality inspection method based on infrared imaging according to claim 5, characterized in that, The calculation of the gray-scale value gradient of the infrared image also includes: performing optimization, that is, taking the square of the components of the difference along the X and Y directions. The calculation is as follows: G(x,y)=[f(x,y)-m(x,y)] 2 +[f(x,y)-n(x,y)] 2 =|gradf(x,y)| 2 7. The PCB board soldering quality inspection method based on infrared imaging according to claim 3, characterized in that, Obtain the pin edge gray-scale gradient function based on the edge point sequence, the gray-scale value gradient calculation of the infrared image, the thermal gradient parameter extraction and the extraction of the gradient function. Its expression is as follows: G(x i ,y i )=|gradP(i)| 2 =|gradf(x i ,y i )| 2 8. The PCB board soldering quality inspection method based on infrared imaging according to claim 2, characterized in that, Identifying potential defect areas existing in the pin solder joints of the PCB board according to the pin edge gray-scale gradient function includes: Using the sequence of boundary points as the x-axis and their corresponding gray-level gradients as the y-axis, a boundary gradient function is established. This function contains local minima and local maxima. When this local minimum is less than a certain threshold V... min Or this local maximum value is greater than a certain threshold V. max At this time, defects will appear in the regions where local minimum and local maximum values ​​are located, and the threshold V min and V max This needs to be obtained through specific experiments; since fluctuations in the gradient function increase the difficulty of extracting some feature parameters, multinomial fitting is used to smooth the original gradient function, and the specific calculation is as follows: T(x)=t1x n +t2x n-1 +t3x n-2 +,...,+t n-1 x 2 +t n x 1 。 9. The PCB board soldering time control method based on infrared imaging according to claim 1, characterized in that, The pin solder joint is: Use the maximum inter-class variance method to perform image segmentation on the original image. When the pixel f(x,y) in the image ≥ T, determine that this point is a pin solder joint. When a certain pixel point f(x,y) < T, let f(x,y) = 0. This point appears black in the image and is classified as a non-pin solder joint; T is the set threshold, and f(x,y) is the original image. The specific calculation for obtaining the segmentation threshold T of the original image using the maximum inter-class variance method is as follows: Assume the original image has L gray levels, and gray level i has n pixel levels. i MN represents the total number of pixels. Given any pixel g(x,y), the probability p that the pixel has a gray level of i is: Assume that the image threshold T(k) used for image segmentation is k, where 0 < k < L-1. This threshold divides the image into two parts, C1 and C2. C1 represents the background, with a gray level in the range [0, k]; C2 represents the target to be extracted, with a gray level in the range [k+1, L-1]. Given an arbitrary pixel, the probability P that this pixel belongs to the background is... A for: Take any pixel, and let P be the probability that this pixel belongs to the target to be extracted. B for: The average gray value ω of parts C1 and C2 can be obtained using the above formula. A and ω B for: Where ω0 is the global grayscale mean, after obtaining ω A ω B After ω0, the inter-class variance σ can be calculated. 2 : s 2 =p A (ω0-ω A ) 2 +p B (oh B -ω0) 2 After calculating the segmentation threshold, perform the corresponding threshold segmentation on the infrared image. The specific calculation is as follows: When the pixel f(x, y) in the image is ≥ T, it is determined that this point is a pin solder joint; when the pixel f(x, y) in the image is < T, it is determined that this point is a non-pin solder joint. Calculate the number of pixels S in each target area, and for the corresponding pin solder joint area, obtain the range of pixel counts (S) of high-quality solder joint areas through experiments. min ,S max ), when S min At this time, appropriately slow down the conveyor rail speed and increase the PCB board immersion time. When S>S max At the same time, the speed of the conveyor rail should be increased appropriately to reduce the PCB board immersion time.​ 10. A PCB board soldering time control method based on infrared imaging, characterized in that, It includes: Obtaining an infrared image of the non-welding side of the PCB board; Processing and analyzing the infrared image to obtain the spatial temperature distribution signal of the non-welding side of the PCB board; Generating a feedback signal according to the temperature signal of the non-welding side of the PCB board, transmitting the feedback signal to the computer, and the computer adjusts the rotational speed of the motor controlling the conveyor rail to achieve the control of the welding time of the wave soldering.

Images