X-ray chip cavity detection method and system based on VIBE algorithm

Abstract

The invention discloses an X-ray chip cavity detection method and system based on a VIBE algorithm. The method comprises the following steps: constructing a background model according to a cavity analysis algorithm; collecting a sequence image of a chip to be detected, and initializing the background model by using a front preset frame number of the sequence image; determining an adaptive segmentation threshold according to the image background complexity, and performing foreground detection and segmentation according to the adaptive segmentation threshold; adjusting the dynamic update rate of the background model according to the image background complexity, and updating the background model according to the dynamic update rate; performing morphological closing operation on the obtained foreground binary image; then connected component analysis is carried out, statistics is carried out on the total pixel number of all connected areas, the ratio of the total pixel number to the total pixel number of the chip to be detected is calculated, and the hole proportion is obtained; according to the invention, the background model of each pixel point can be established by analyzing the image of the chip area. And then, comparing each pixel of the current image with the background model so as to separate the foreground in the image.

Description

Technical Field

[0001] This invention relates to the field of integrated circuit design technology, and in particular to a method and system for detecting voids in X-ray chips based on the VIBE algorithm. Background Technology

[0002] In X-ray inspection of the semiconductor industry, accurately identifying voids and defects inside chip packages is a crucial step in ensuring product quality. Traditional void detection methods often employ simple threshold segmentation, edge detection, or static background subtraction. These methods are effective for scenarios with uniform illumination and a static background, but in practical applications, X-ray imaging is subject to noise, brightness fluctuations, micro-motion of the chip, and textural interference from the packaging material itself, resulting in high false alarm rates and poor adaptability of traditional methods.

[0003] The Visual Background Extractor (VIBE) algorithm is a high-efficiency pixel-level background modeling algorithm. It quickly initializes and adapts to scene changes by building a sample set for each pixel and employing a random update strategy. Its basic process includes: 1) Model initialization: Randomly sampling using spatial neighborhood information from the first frame image to build a background model for each pixel; 2) Foreground detection: Comparing the current frame pixel with the background model. If a sufficient number of sample points are less than a fixed threshold away from the current pixel, it is identified as background; otherwise, it is identified as foreground; 3) Model update: For pixels identified as background, randomly updating their own and neighboring pixel models with a certain probability.

[0004] However, directly applying the traditional VIBE algorithm to chip hole detection faces many problems: Motion blur and ghosting issues: When X-ray inspection equipment captures images of different chips, the movement of the camera introduces motion blur and ghosting. When the first frame image is blurry or has ghosting, this motion is directly learned into the background model. This makes the established background model itself inaccurate and unclear, severely affecting the foreground detection accuracy of all subsequent frames, causing blurred areas to be misclassified as background or foreground.

[0005] Limitations of fixed threshold: Using a globally fixed segmentation threshold R cannot adapt to local variations in different regions of the image. For chip regions with complex textures and large contrast differences, a fixed threshold can lead to missed detections in simple regions and false detections in complex regions.

[0006] Fixed update rate drawbacks: Using a fixed update probability φ, it cannot adaptively adjust according to the drastic changes in the scene. Updating too quickly when the background is stable will introduce noise, while updating too slowly when the scene changes abruptly (such as changes in lighting) will cause the algorithm to lag and produce motion blur.

[0007] These shortcomings severely limit the application of the traditional VIBE algorithm in high-precision, high-reliability industrial inspection. Summary of the Invention

[0008] To address the shortcomings of the existing technologies, the technical problem this invention aims to solve is to propose a method and system for detecting X-ray chip voids based on the VIBE algorithm. This method can establish a background model for each pixel by analyzing the image of the chip region. Then, each pixel of the current image is compared with the background model to separate the foreground (void region) in the image. Finally, if a pixel is determined to be a background point, its background model is updated to maintain the accuracy and effectiveness of the model.

[0009] One technical solution adopted in this invention is: a method for detecting voids in X-ray chips based on the VIBE algorithm, which includes the following steps: S1: Construct a background model based on the void analysis algorithm; S2: Acquire a sequence of images of the chip under test, and initialize the background model using the preset number of frames in the sequence of images; S3: Determine the adaptive segmentation threshold based on the image background complexity, and perform foreground detection and segmentation based on the adaptive segmentation threshold; S4: Adjust the dynamic update rate of the background model according to the complexity of the image background, and update the background model according to the dynamic update rate; S5: Perform morphological closing operation on the obtained foreground binary image; then perform connected component analysis, count the total number of pixels in all connected regions, calculate the ratio of the total number of pixels to the total number of pixels in the chip under test, and obtain the hole ratio.

[0010] Furthermore, the background model in step S1 is the set of pixels in a preset neighborhood space of any pixel in the sequence of images, specifically represented as follows: ; in, Represents the number of pixels in the t-th frame of the image. Background model, This indicates the pixel point The neighboring pixel values, where m represents the number of sample points.

[0011] Furthermore, step S2 includes the following sub-steps: S21: Acquire the sequence of images output by the X-ray detection equipment, and take the previous preset number of frames as the initialization source of the background model; S22: Initialize the background model based on the previously preset number of frames: ; in, Represents the number of pixels in the first N frames of the image. Initialize the background model. This represents the number of pixels in the f-th frame of the image. The neighboring pixel values, M represents the number of frames and M represents the number of samples.

[0012] Furthermore, step S3 includes the following sub-steps: S31: For each pixel in the current frame, calculate the standard deviation of the gray values ​​in its local neighborhood space; S32: Calculate the adaptive segmentation threshold based on the standard deviation of gray values; S33: Calculate the Euclidean distance between each pixel and each sample point in the background model of that pixel, and count the number of sample points whose Euclidean distance is greater than or equal to the adaptive segmentation threshold. If the number is greater than or equal to the preset comparison threshold, then the pixel is the background.

[0013] Furthermore, the specific calculation process of step S31 is as follows: ; ; in, Represents pixels The local neighborhood space, Represents pixels The mean of the local neighborhood space, Represents pixels The standard deviation of gray values ​​in the local neighborhood space.

[0014] Furthermore, the specific calculation process of step S32 is as follows: ; in, Indicates the adaptive segmentation threshold. Indicates a fixed segmentation threshold; This represents the minimum threshold of standard deviation. represents the maximum standard deviation threshold, used to classify background complexity; a and b represent control parameters used to adjust the segmentation threshold.

[0015] Furthermore, step S4 includes the following sub-steps: S41: For pixels identified as background, adjust the update rate based on the standard deviation of the pixel's grayscale value; S42: Update the background model for the corresponding pixels based on the adjusted update rate.

[0016] Furthermore, the specific calculation process of step S41 is as follows: ; in, Indicates the update rate. Indicates the original fixed update rate. , This represents a control parameter used to adjust the update rate.

[0017] The second technical solution adopted in this invention is an X-ray chip cavity detection system based on the VIBE algorithm, which includes the following modules: The background model building module is used to build a background model based on the void analysis algorithm. The initialization module is used to acquire a sequence of images of the chip under test and initialize the background model using the preset number of frames of the sequence images. The segmentation module is used to determine an adaptive segmentation threshold based on the complexity of the image background, and to perform foreground detection and segmentation based on the adaptive segmentation threshold. The model update module is used to adjust the dynamic update rate of the background model according to the complexity of the image background, and update the background model according to the dynamic update rate. The hole calculation module is used to perform morphological closing operations on the obtained foreground binary image; then, connected component analysis is performed to count the total number of pixels in all connected regions and calculate the ratio of the total number of pixels to the total number of pixels in the chip under test to obtain the hole ratio.

[0018] Furthermore, the initialization module includes the following sub-modules: The image acquisition submodule is used to acquire the sequence of images output by the X-ray detection equipment and take the previous preset number of frames as the initialization source of the background model; The multi-frame initialization submodule is used to initialize the background model based on the previous preset number of frames.

[0019] The present invention provides a method and system for detecting voids in X-ray chips based on the VIBE algorithm, which has at least the following beneficial effects: 1. High-precision detection: Motion blur and ghosting are eliminated through multi-frame initialization, and segmentation accuracy is optimized through adaptive thresholding, which greatly improves the recall and precision of hole detection.

[0020] 2. Strong robustness: The adaptive threshold and dynamic update mechanism enable the algorithm to effectively cope with common noise, illumination changes and background texture interference in X-ray images, and the stability is significantly enhanced.

[0021] 3. High efficiency: The algorithm retains the high computational efficiency of the VIBE algorithm and is suitable for industrial online real-time detection systems.

[0022] 4. Practicality: The final output provides a quantitative result of the void ratio and a pass / fail judgment based on industry standards, which is intuitive and effective, greatly improving the automation level and practical value of the test. Attached Figure Description

[0023] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of an X-ray chip cavity detection method based on the VIBE algorithm according to the present invention.

[0024] Figure 2 for Figure 1 The sub-flowchart of step S2.

[0025] Figure 3 for Figure 1 The sub-flowchart of step S3 in the middle.

[0026] Figure 4 for Figure 1 The sub-flowchart of step S4 in the middle.

[0027] Figure 5 The image shows the analysis results of a chip cavity analysis using an X-ray chip cavity detection method based on the VIBE algorithm of this invention.

[0028] Figure 6 This is a block diagram of an X-ray chip cavity detection system based on the VIBE algorithm according to the present invention. Detailed Implementation

[0029] The invention will now be further described with reference to the accompanying drawings.

[0030] The traditional VIBE algorithm, when directly applied to chip hole detection, generally includes the following steps: Model initialization: The hole analysis algorithm initializes the background model using the first frame of the image sequence. Based on the similar spatial distribution characteristics of adjacent pixels, it randomly selects neighboring pixels as sample values ​​for the background model, utilizing the spatial consistency of pixels. For any pixel... Establish a background model consisting of m sample points. Its neighboring pixels are Then, the background model was initialized by randomly selecting neighboring pixels m times. This random method ensured the objectivity and impartiality of the model and eliminated subjective bias factors. Represents the pixel at time t pixel values, express Moment Pixel The neighboring pixel values, express Moment Pixel The background model, then The definition of is:

[0031] Foreground detection: After the background model is initialized, each pixel of the current frame is compared with the sample points in the model. First, the relationship between each pixel and its model is calculated. The Euclidean distance of each sample point is calculated, and then the number of sample points whose Euclidean distance is less than a segmentation threshold is counted. If the number of such samples is greater than a pre-set threshold, the pixel is determined to belong to the background (BG); otherwise, it is determined to be a foreground (FG). The specific determination method is as follows:

[0032] in, This represents the Euclidean distance between a pixel and its background model sample points. This indicates the number of sample points that meet the conditions. The segmentation threshold is... This is a threshold for the number of matches. If the distance between the current pixel and the background model sample point is less than... The number of items is greater than or equal to If it is true, it is classified as BG; otherwise, it is classified as FG.

[0033] Background model update: After foreground detection, if a pixel is identified as a background point, its background model needs to be updated. The VIBE algorithm's update strategy mainly follows these principles.

[0034] 1. To ensure that the survival time of each sample point in the model decays exponentially, the current pixel is randomly replaced. One of the sample points:

[0035] 2. Unlike other update methods, the VIBE algorithm is based on... The probability of updating the background model, where, This is the time sampling factor, also known as the original fixed update rate.

[0036] 3. While updating the background point, the hole analysis algorithm also randomly selects the background model of the neighboring pixels of that point and updates it according to steps 1 and 2. The neighborhood update strategy can reduce the impact of camera shake and slight target movement during shooting, thus improving detection accuracy.

[0037] 4. If a pixel is identified as a foreground point for multiple consecutive frames, then update it as a background point.

[0038] Please see Figure 1 The diagram below illustrates a method for detecting voids in X-ray chips based on the VIBE algorithm, according to the present invention. This method may include the following steps: S1: Construct a background model based on the void analysis algorithm. The method for constructing the background model in step S1 is the same as the traditional method described above.

[0039] Specifically, the background model in step S1 is the set of pixels in a preset neighborhood space of any pixel in the sequence of images. This background model is specifically represented as follows: ; in, Represents the number of pixels in the t-th frame of the image. Background model, This indicates the pixel point The neighboring pixel values, where m represents the number of sample points.

[0040] S2: Acquire a sequence of images of the chip under test and initialize the background model using the preset number of previous frames. The traditional hole analysis algorithm described above initializes the background model with a single frame. Its advantage lies in shortening the model initialization time. Furthermore, if a significant change in the scene is detected, model initialization can be completed again simply by reusing the changed first frame. However, moving objects often exist in the first frame of the image sequence. The VIBE algorithm uses pixel spatial consistency to initialize the background model, which may learn pixels that originally belonged to the foreground into the background model. At the start of detection, foreground points are mistakenly identified as background points, resulting in motion blur and ghosting. It takes time to update the background model to eliminate these issues. Therefore, this algorithm is only suitable for scenes where the background is initially "static." If the background is "moving," it will severely affect the detection results. In summary, since the spatial consistency of a pixel within a single frame cannot accurately determine whether a pixel is a foreground point, step S2 uses the pixel values ​​of consecutive previous frames to initialize the background model. This completely solves the motion blur and ghosting problems in the early stages of detection, resulting in a "clean" background. Please see Figure 2 This is a sub-flowchart for step S2, which may include the following sub-steps: S21: Acquire the sequence of images output by the X-ray detection equipment, and take the previous preset number of frames as the initialization source of the background model; S22: Initialize the background model based on the previously preset number of frames: ; in, Represents the number of pixels in the first N frames of the image. Initialize the background model. This represents the number of pixels in the f-th frame of the image. The neighboring pixel values, Here, S represents the frame number, and M represents the number of samples. Steps S21-S22 are an improved algorithm that initializes the background model based on the previous N frames.

[0041] S3: Determine an adaptive segmentation threshold based on the image background complexity, and perform foreground detection and segmentation based on the adaptive segmentation threshold. In the foreground detection stage, traditional hole analysis algorithms use a global threshold for segmentation. However, the background is not static, and this method is not ideal for complex situations such as sudden changes in lighting and dynamic backgrounds, resulting in a large number of false detections and alarms. The improved hole analysis algorithm uses the background complexity obtained from pixel spatial neighborhood information to determine the segmentation threshold, reducing the impact of dynamic backgrounds on detection to some extent. Analysis shows that when the scene is complex, pixels may be misclassified as foreground, so the segmentation threshold should be larger; when the scene is simple, even subtle changes are caused by the foreground, so the threshold should be smaller to ensure detection accuracy. The background complexity in step S3 is determined by calculating the standard deviation within the region where the pixel is located. The standard deviation reflects the degree of deviation of a pixel from the average value of that region. A larger standard deviation indicates greater differences between pixels in that region, a more complex background, and therefore a larger segmentation threshold; conversely, a smaller standard deviation indicates less difference between pixels in that region, a simpler background, and therefore a smaller segmentation threshold.

[0042] Please see Figure 3 This is a sub-flowchart of step S3. Step S3 may also include the following sub-steps: S31: For each pixel in the current frame, calculate the standard deviation of the gray values ​​in its local neighborhood space; Specifically, the calculation process for step S31 is as follows: ; ; in, Represents pixels The local neighborhood space, Represents pixels The mean of the local neighborhood space, Represents pixels Standard deviation of gray values ​​in the local neighborhood space S32: Calculate the adaptive segmentation threshold based on the standard deviation of gray values; Specifically, the calculation process for step S32 is as follows: ; in, Indicates the adaptive segmentation threshold. Indicates a fixed segmentation threshold; This represents the minimum threshold of standard deviation. represents the maximum standard deviation threshold, used to classify background complexity; a and b represent control parameters used to adjust the segmentation threshold.

[0043] S33: Calculate the Euclidean distance between each pixel and each sample point in the background model of that pixel, and count the number of sample points whose Euclidean distance is greater than or equal to the adaptive segmentation threshold. If the number is greater than or equal to the preset comparison threshold, then the pixel is the background.

[0044] S4: Adjust the dynamic update rate of the background model based on the complexity of the image background, and update the background model according to the dynamic update rate. Traditional hole analysis algorithms randomly update sample points in the background model, and simultaneously update neighboring pixels in the same way, but at a fixed update rate. In reality, updates that are too fast or too slow cannot adapt well to scene changes. Updates that are too fast lead to incomplete detection, resulting in missed detections and holes appearing in the target; updates that are too slow result in motion blur, especially when the scene changes abruptly, reducing detection quality. The improved hole analysis algorithm uses background complexity to achieve dynamic updates. The more complex the background, the higher the probability of false detections, so the update speed should be slowed down. At the same time, to avoid the spread of erroneous pixels, neighboring updates should be suppressed. Conversely, when the background is simple, the update speed should be increased.

[0045] Please see Figure 4 This is a sub-flowchart for step S4. Step S4 may also include the following sub-steps: S41: For pixels identified as background, adjust the update rate based on the standard deviation of the pixel's grayscale value; Specifically, the calculation process of step S41 is as follows: ; in, Indicates the update rate. Indicates the original fixed update rate. , This represents a control parameter used to adjust the update rate.

[0046] S42: Update the background model for the corresponding pixels based on the adjusted update rate.

[0047] S5: Perform morphological closing operation on the obtained foreground binary image; then perform connected component analysis, count the total number of pixels in all connected regions, calculate the ratio of the total number of pixels to the total number of pixels in the chip under test, and obtain the hole ratio.

[0048] Please see Figure 5 The image shows the analysis results of a chip cavity analysis using an X-ray chip cavity detection method based on the VIBE algorithm of this invention. The cavity analysis process specifically includes the following steps: 1. System Preparation: Deploy a computing device equipped with the algorithm of this invention and connect it to an X-Ray imaging device. Start the "X-Ray Chip Intelligent Hole Defect Discrimination System" software, and load the pre-trained chip localization YOLO model and configuration parameters.

[0049] 2. Image Acquisition and Input: The sequence of images of the chip under test is acquired through an X-Ray device and transmitted to the system image input module.

[0050] 3. Chip Region Localization: The YOLO model is used to infer the bounding box coordinates (x1, y1, x2, y2) of all chips in the image. Each chip region is then cropped as the region of interest (ROI) for subsequent hole analysis.

[0051] 4. Improved VIBE algorithm execution: Select preset parameters and execute the improved VIBE algorithm independently on each chip ROI image sequence to obtain a foreground binary image.

[0052] 5. Post-processing and hole analysis: Morphological closing operations are performed on the obtained foreground binary image to fill small holes and smooth the boundaries. Then, connected component analysis is performed to count the total number of pixels in all connected regions and calculate the ratio of this ratio to the total number of pixels in the chip's ROI to obtain the hole percentage.

[0053] In summary, compared with the traditional detection algorithms mentioned above, this solution has the following advantages: High-precision detection: Motion blur and ghosting are eliminated through multi-frame initialization, and segmentation accuracy is optimized through adaptive thresholding, which greatly improves the recall and precision of hole detection.

[0054] Strong robustness: The adaptive threshold and dynamic update mechanism enable the algorithm to effectively cope with common noise, illumination changes and background texture interference in X-ray images, and the stability is significantly enhanced.

[0055] High efficiency: The algorithm retains the high computational efficiency of the VIBE algorithm, making it suitable for industrial online real-time detection systems.

[0056] Practicality: The final output provides a quantitative result of the void ratio and a pass / fail judgment based on industry standards, which is intuitive and effective, greatly improving the automation level and practical value of the test.

[0057] Please see Figure 6The diagram shows a structural block diagram of an X-ray chip cavity detection system based on the VIBE algorithm according to the present invention. The present invention also provides an X-ray chip cavity detection system based on the VIBE algorithm, which is used to implement the aforementioned X-ray chip cavity detection method based on the VIBE algorithm. Specifically, the system may include a background model construction module 100, an initialization module 200, a segmentation module 300, a model update module 400, and a cavity calculation module 500. Specifically: Background model construction module 100 is used to construct a background model based on the void analysis algorithm; The initialization module 200 is used to acquire a sequence of images of the chip under test and initialize the background model using the preset number of frames of the sequence of images. The initialization module 200 includes the following sub-modules: The image acquisition submodule is used to acquire the sequence of images output by the X-ray detection equipment and take the previous preset number of frames as the initialization source of the background model; The multi-frame initialization submodule is used to initialize the background model based on the previous preset number of frames. The segmentation module 300 is used to determine an adaptive segmentation threshold based on the complexity of the image background, and to perform foreground detection and segmentation based on the adaptive segmentation threshold. The model update module 400 is used to adjust the dynamic update rate of the background model according to the complexity of the image background, and update the background model according to the dynamic update rate. The hole calculation module 500 is used to perform morphological closing operations on the obtained foreground binary image; then, connected component analysis is performed to count the total number of pixels in all connected regions and calculate the ratio of the total number of pixels to the total number of pixels in the chip under test to obtain the hole ratio.

[0058] The above description merely illustrates preferred embodiments of the present invention and is quite specific and detailed; however, it should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of the present invention. Therefore, the scope of protection of this invention should be determined by the appended claims.

Claims

1. An X-ray chip cavity detection method based on a VIBE algorithm, comprising the following steps: S1: constructing a background model according to a cavity analysis algorithm; S2: collecting a sequence image of a chip to be detected, and initializing the background model by using a preset number of initial frames of the sequence image; S3: determining an adaptive segmentation threshold according to the background complexity of the image, and performing foreground detection and segmentation according to the adaptive segmentation threshold; S4: adjusting the dynamic update rate of the background model according to the background complexity of the image, and updating the background model according to the dynamic update rate; S5: performing a morphological closing operation on the obtained foreground binary image; then performing connected component analysis, counting the total number of pixels of all connected regions, calculating the ratio of the total number of pixels to the total number of pixels of the chip to be detected, and obtaining the cavity ratio. 2.The X-ray chip void detection method based on VIBE algorithm of claim 1, wherein, The background model in the S1 step is a pixel point set of a preset neighborhood space of any pixel point in any frame of the sequence image, and the background model is specifically represented as: ； wherein, denotes the background model of pixel point in the t-th frame image, denotes the neighborhood pixel values of the pixel point , and m denotes the number of sample points. 3.The X-ray chip void detection method based on VIBE algorithm of claim 1, wherein, The S2 step comprises the following sub-steps: S21: collecting a sequence image output by an X-ray detection device, and taking a preset number of initial frames of images as the initialization source of the background model; S22: initializing the background model according to the preset number of initial frames of images: ； wherein, represents an initialized background model of pixel point in the first N frames of images, represents the neighborhood pixel value of pixel point in the fth frame of images, represents the frame number, and M represents the sample number. 4.The X-ray chip void detection method based on VIBE algorithm of claim 1, wherein, The S3 step comprises the following sub-steps: S31: calculating the standard deviation of the gray value of the local neighborhood space of each pixel point in the current frame; S32: calculating the adaptive segmentation threshold according to the standard deviation of the gray value; S33: calculating the Euclidean distance between each pixel point and each sample point in the background model of the pixel point, and counting the number of sample points whose Euclidean distance is greater than or equal to the adaptive segmentation threshold, and if the number is greater than or equal to a preset comparison threshold, the pixel point is background.

5. The X-ray chip void detection method and system based on VIBE algorithm of claim 4, wherein, The specific calculation process of the S31 step is: ； ； wherein, denotes the local neighborhood space of a pixel point , denotes the mean value of the local neighborhood space of a pixel point , denotes the standard deviation of the gray value of the local neighborhood space of a pixel point .

6. The X-ray chip void detection method based on VIBE algorithm of claim 4, wherein, The specific calculation process of the S32 step is: ； wherein, represents an adaptive segmentation threshold, represents a fixed segmentation threshold; represents a standard deviation minimum threshold, represents a standard deviation maximum threshold for classifying background complexity; a, b represent control parameters for adjusting the segmentation threshold. 7.The X-ray chip void detection method based on VIBE algorithm of claim 4, wherein, The S4 step comprises the following sub-steps: S41: adjusting the update rate of the pixel point according to the standard deviation of the gray value of the pixel point; S42: updating the background model of the corresponding pixel point according to the adjusted update rate. 8.The X-ray chip void detection method based on VIBE algorithm of claim 7, wherein, The specific calculation process of the S41 step is: ； wherein, represents an update rate, represents an original fixed update rate, , represents a control parameter for adjusting the increase or decrease of the update rate.

9. An X-ray chip void detection system based on VIBE algorithm, characterized in that, The system comprises the following modules: a background model construction module for constructing a background model according to a cavity analysis algorithm; an initialization module for collecting a sequence image of a chip to be detected, and initializing the background model by using a preset number of initial frames of the sequence image; a segmentation module for determining an adaptive segmentation threshold according to the background complexity of the image, and performing foreground detection and segmentation according to the adaptive segmentation threshold; a model update module for adjusting the dynamic update rate of the background model according to the background complexity of the image, and updating the background model according to the dynamic update rate; a cavity calculation module for performing a morphological closing operation on the obtained foreground binary image; then performing connected component analysis, counting the total number of pixels of all connected regions, calculating the ratio of the total number of pixels to the total number of pixels of the chip to be detected, and obtaining the cavity ratio.

10. The X-ray chip void detection system based on VIBE algorithm as claimed in claim 9, wherein, The initialization module comprises the following sub-modules: an image acquisition sub-module for collecting a sequence image output by an X-ray detection device, and taking a preset number of initial frames of images as the initialization source of the background model; a multi-frame initialization sub-module for initializing the background model according to the preset number of initial frames of images.

Images