Chip defect detector based on multispectral imaging

Abstract

The invention relates to the technical field of chip detection, in particular to a chip defect detector based on multispectral imaging, which comprises a detector main body, a discharge port is arranged at the center of the front surface of the detector main body, an adjustable fixing mechanism is arranged in the detector main body, and an image taking mechanism is arranged above the fixing mechanism. According to the invention, by optimizing a fixing and image capturing structure and combining a multispectral imaging and intelligent analysis system, all-directional dead-corner-free detection of the chip is realized. The multi-dimensional adjustment capability of the fixing mechanism guarantees the detection stability and comprehensiveness, and the spectrum switching and light path design of the image capturing mechanism improves the defect identification precision. The equipment integrates an automatic process and an intelligent analysis function, greatly reduces the omission ratio and manual intervention, is compatible with chips of various specifications, gives consideration to the detection efficiency and the refinement level, provides an efficient and accurate solution for chip manufacturing quality management and control, and assists in improving the reliability and market competitiveness of chip products.

Description

Technical Field

[0001] This invention relates to the technical field of chip inspection, specifically to a chip defect detector based on multispectral imaging. Background Technology

[0002] As a core component of the electronic information industry, chips undergo increasingly sophisticated manufacturing processes, with even minute defects directly determining product reliability and lifespan. Traditional chip inspection technologies rely heavily on single-spectrum imaging or manual screening, which has significant limitations: single-spectrum imaging struggles to capture the optical response differences between different types of defects, leading to high rates of missed detections of minute cracks and internal packaging defects; manual inspection is inefficient and highly subjective, failing to meet the quality control requirements of large-scale mass production. Existing inspection equipment suffers from shortcomings in clamping stability and multi-faceted inspection compatibility, making it difficult to achieve comprehensive, blind-spot-free inspection of the chip's front and back surfaces and sides. Lagging spectral switching response further restricts the level of precision in quality control. Therefore, there is an urgent need for an inspection device that integrates efficient clamping and precise multi-spectral imaging. Summary of the Invention

[0003] The purpose of this invention is to provide a chip defect detector based on multispectral imaging to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a chip defect detector based on multispectral imaging, comprising a detector body, a feeding port provided at the center of the front of the detector body, an adjustable fixing mechanism provided inside the detector body for fixing the chip to be inspected, and an image acquisition mechanism provided above the fixing mechanism for acquiring images of the chip to be inspected under different light irradiation conditions.

[0005] Preferably, the discharge port is a square channel, a sealing cover is hinged at the open end of the square channel, and a conveyor belt is installed inside the square channel.

[0006] Preferably, the fixing mechanism includes a mounting base, which is located in the middle of the bottom surface of the detector body. A first telescopic rod is installed inside the mounting base, a chip tray is provided on the top of the first telescopic rod, and clamping components are provided on both sides of the chip tray.

[0007] Preferably, the first telescopic rod and the discharge port are aligned.

[0008] Preferably, the clamping assembly includes a clamping block, one end of which is fixedly connected to a second telescopic rod. The second telescopic rod is rotatably mounted on a rotating mounting block, which is located on the inner side wall of the main body of the detector. A third adjusting motor is installed inside the rotating mounting block to control the rotation of the second telescopic rod. The other end of the clamping block has a notch, and a square groove is formed on the inner top surface of the notch. A lifting clamping plate is movably disposed in the square groove. A clamping push rod is installed on the top of the clamping block, one end of which passes through the upper part of the clamping block and is connected to the lifting clamping plate. A drive wheel is embedded in the inner bottom surface of the notch.

[0009] Preferably, two ball bearings are embedded in the bottom surface of the lifting clamping plate, a rotating shaft is provided at the center of the drive wheel, the rotating shaft is rotatably installed in the clamping block, a first connecting wheel is provided at one end of the rotating shaft, the lower part of the clamping block is a hollow structure, a second adjusting motor is provided inside the lower part of the clamping block, the output end of the second adjusting motor is connected to the second connecting wheel, and the second connecting wheel and the first connecting wheel are connected by a synchronous belt.

[0010] Preferably, the imaging mechanism includes a mounting top plate, which is disposed on the inner top surface of the detector body. An imaging device is disposed at the center of the bottom surface of the mounting top plate, and several sets of adjustable light source components are disposed around the outer side of the imaging device.

[0011] Preferably, the light source assembly includes a light source mounting plate, a cylindrical mounting block is provided on the bottom surface of the mounting top plate, the light source mounting plate is rotatably mounted on the cylindrical block by an inclined metal rod, an illumination light source is mounted on the bottom surface of the light source mounting plate, an external toothed ring is provided on the side wall of the light source mounting plate, an inner groove is provided on the bottom surface of the mounting top plate, an inner rotating ring is rotatably mounted in the inner groove, a bottom ring is connected below the inner rotating ring, an internal toothed ring is provided on the inner side wall of the bottom ring, the internal toothed ring and the external toothed ring are meshed together, the upper end of the inner side wall of the inner rotating ring is a sloped structure, an arc-shaped rack is provided on the sloped surface, a first adjusting motor is mounted in the mounting top plate, the output end of the first adjusting motor is connected to an adjusting gear, the adjusting gear and the arc-shaped rack are meshed together.

[0012] Preferably, there are three illumination light sources, which respectively illuminate ultraviolet light, visible light and near-infrared light.

[0013] Preferably, the chip defect detector further includes a defect detection system, the defect detection system comprising: The loading and unloading module is used to load and unload chips. Loading is the accurate transfer of chips from the conveyor belt in the unloading port to the chip tray, and unloading is the transfer of the tested chips from the chip tray to the opening of the unloading port. The chip fixing module is used to control the fixing mechanism to fix and clamp the chip and adjust its state. The image acquisition module captures images of the chip using an imaging device and converts the light signal into a 16-bit grayscale digital image. After acquiring images of all preset bands, it forms a multispectral image set (ultraviolet, visible, and near-infrared images) and transmits it to the image preprocessing module in real time. The image acquisition module also includes a light source control unit, which controls the switching on and off of the light source, brightness adjustment, and switching between different light sources. The image preprocessing module, including an FPGA image accelerator, is used for real-time data processing. It eliminates image noise, corrects positional offsets, and amplifies defect features through algorithms, ensuring the quality of the data input to the analysis module. The core algorithm principle is as follows: Noise suppression: Median filtering is used to preserve the details of defect edges and avoid excessive smoothing; Image registration: Based on the phase correlation method, the coordinates of images in different bands are aligned to ensure that the same defect is in the same position in images of different bands; Contrast enhancement: Adaptive histogram equalization is used to amplify the grayscale difference between defective and normal areas; The analysis module filters suspected defects, extracts features, judges defects, quantifies their size and level, and finally outputs clear defect information from the preprocessed multispectral image set. The data storage and interaction module is used to display the inspection progress, defect images, and inspection reports in real time, supports report export, and triggers audible and visual alarms for serious defects.

[0014] Compared with the prior art, the beneficial effects of the present invention are: This invention achieves omnidirectional, blind-spot-free chip inspection by optimizing the fixation and imaging structures and combining them with a multispectral imaging and intelligent analysis system. The multidimensional adjustment capability of the fixation mechanism ensures the stability and comprehensiveness of the inspection, while the spectral switching and optical path design of the imaging mechanism improves the accuracy of defect identification. The equipment integrates automated processes and intelligent analysis functions, significantly reducing the false negative rate and manual intervention. It is compatible with various chip specifications, balancing inspection efficiency and precision, providing an efficient and accurate solution for chip manufacturing quality control, and helping to improve the reliability and market competitiveness of chip products. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the device structure of the present invention.

[0016] Figure 2 This is a schematic diagram of the opening structure of the discharge port 2 of the present invention.

[0017] Figure 3 This is a schematic diagram of the internal structure of the main body 1 of the detector of the present invention.

[0018] Figure 4 This is a schematic diagram of the fixing mechanism of the present invention.

[0019] Figure 5 This is a schematic diagram of the clamping component structure of the present invention.

[0020] Figure 6 This is a schematic diagram of the clamping component of the present invention from another perspective.

[0021] Figure 7 This is a schematic diagram of the connection structure of the drive wheel 22 of the present invention.

[0022] Figure 8 This is a schematic diagram of the structure of the mounting plate 3 of the present invention.

[0023] Figure 9 This is a schematic diagram of the light source component structure of the present invention.

[0024] Figure 10 This is a schematic diagram of the connection structure of the bottom ring 8 of the present invention.

[0025] Figure 11 This is a flowchart of the system structure of the present invention.

[0026] In the diagram: 1. Detector body; 2. Feed port; 3. Mounting top plate; 4. Light source mounting plate; 5. Irradiation light source; 6. External gear ring; 7. Imaging device; 8. Bottom ring; 9. Internal gear ring; 10. Internal rotating ring; 11. Arc-shaped rack; 12. First adjusting motor; 13. Adjusting gear; 14. Mounting base; 15. First telescopic rod; 16. Chip support plate; 17. Clamping block; 18. Second telescopic rod; 19. Rotating mounting block; 20. Clamping push rod; 21. Lifting clamping plate; 22. Drive wheel; 23. Rotating shaft; 24. First connecting wheel; 25. Second adjusting motor; 26. Second connecting wheel. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Please see Figures 1 to 11 The present invention provides a technical solution: a chip defect detector based on multispectral imaging, including a detector body 1, wherein a feeding port 2 is provided at the center of the front of the detector body 1, the feeding port 2 is a square channel, a sealing cover plate is hinged at the opening end of the square channel, and a conveyor belt is provided inside the square channel.

[0029] The detector body 1 is equipped with an adjustable fixing mechanism inside, which is used to fix the chip to be tested. The fixing mechanism includes a mounting base 14, which is located in the middle of the bottom surface of the detector body 1. A first telescopic rod 15 is installed inside the mounting base 14. The first telescopic rod 15 is aligned with the feeding port 2. A chip tray 16 is provided on the top of the first telescopic rod 15. A structure to reduce friction can be provided on the surface of the chip tray 16, such as a coating or a rolling structure, so as to ensure smoother chip loading and unloading. Clamping components are provided on both sides of the chip tray 16.

[0030] During the feeding process, first open the sealing cover at the opening of the feeding port 2, and then use a robotic arm or other device or manually to clamp the chip to be tested and place it at the designated position on the conveyor belt. The conveyor belt starts to transport the chip into the body 1 of the testing instrument. At this time, the chip tray 16 is located at the other end of the conveyor belt. The chip tray 16 is flush with or slightly lower than the height of the conveyor belt. The chip is transferred onto the chip tray 16. Then, the first telescopic rod 15 extends to push the chip tray 16 and the chip above it upwards. After being pushed up to the height of the clamping assembly, the clamping assembly clamps and fixes the chip.

[0031] The clamping assembly includes a clamping block 17, one end of which is fixedly connected to a second telescopic rod 18. The second telescopic rod 18 is rotatably mounted on a rotating mounting block 19, which is located on the inner wall of the main body 1 of the detector. A third adjusting motor is installed inside the rotating mounting block 19 to control the rotation of the second telescopic rod 18. The other end of the clamping block 17 has a notch, and a square groove is formed on the inner top surface of the notch. A lifting clamping plate 21 is movably disposed in the square groove. A clamping push rod 20 is installed on the top of the clamping block 17, one end of which passes through the upper part of the clamping block 17 and is connected to the lifting clamping plate 21. A drive wheel 22 is embedded in the inner bottom surface of the notch. Two ball bearings are embedded in the bottom surface of the lifting clamping plate 21. A rotating shaft 23 is located at the center of the drive wheel 22. Shaft 23 is rotatably mounted in clamping block 17. A first connecting wheel 24 is provided on one end of shaft 23. Clamping block 17 is a structure welded together. The lower part of clamping block 17 is a hollow structure. A second adjusting motor 25 is provided inside the lower part of clamping block 17. The output end of the second adjusting motor 25 is connected to a second connecting wheel 26. The second connecting wheel 26 and the first connecting wheel 24 are connected by a synchronous belt. When clamping and fixing the chip, the second telescopic rod 18 is first controlled to extend, so that the clamping blocks 17 at both ends move towards the middle until the two ends of the chip enter the notch of clamping block 17. Then, the clamping push rod 20 is controlled to extend and push the lifting clamping plate 21 down. The lifting clamping plate 21 and the drive wheel 22 clamp and fix the chip. The chip is fixed at six points by the ball bearings at both ends and the drive wheel 22. The ball bearings and the drive wheel 22 are movable, so that the chip can move in clamping block 17.

[0032] When the imaging mechanism captures images of the chip, it controls the first telescopic rod 15 to retract, moving the chip holder 16 downwards. First, it samples the front side of the chip. Then, it controls the third adjusting motor to start, driving the second telescopic rod 18 to rotate (stepper motors are recommended for precise control via a microcontroller in this solution), causing the chip to flip at a certain angle (e.g., 90° or 180°). This allows for sampling of the sides and back of the chip. The two sides of the chip that are clamped also need to be sampled for testing. At this time, it controls the second adjusting motor 25 to start, driving the second connecting wheel 26 to rotate, thereby causing the drive wheel 2... Driven by the 2nd drive wheel, the two ends of the drive wheel 22 rotate in opposite directions, allowing the chip to rotate between the two clamping blocks 17. When the side clamped in the notch of the clamping block 17 is replaced by the other two sides, all sidewalls of the chip can be sampled and tested. At the clamping position of the clamping block 17, the clamping block 17 itself will also cause obstruction, resulting in incomplete sampling. At this time, it is only necessary to control the second adjustment motor 25 to rotate again to drive the drive wheel 22 to rotate, and control the drive wheels 22 at both ends to rotate in the same direction, so that the chip can be translated until the obstructed part is exposed, and the chip can be fully sampled and tested.

[0033] An image-capturing mechanism is provided above the fixed mechanism. The image-capturing mechanism is used to capture images of the chip to be inspected under different lighting conditions. The image-capturing mechanism includes a mounting top plate 3, which is disposed on the inner top surface of the detector body 1. An imaging device 7 is disposed at the center of the bottom surface of the mounting top plate 3. Several sets of adjustable light source components are disposed around the outer side of the imaging device 7. The light source assembly includes a light source mounting plate 4. A cylindrical mounting block is provided on the bottom surface of the mounting top plate 3. The light source mounting plate 4 is rotatably mounted on the cylindrical block via an inclined metal rod. Irradiation light sources 5 are mounted on the bottom surface of the light source mounting plate 4. There are three irradiation light sources 5, which respectively irradiate ultraviolet light, visible light, and near-infrared light. An external toothed ring 6 is provided on the side wall of the light source mounting plate 4. An inner groove is provided on the bottom surface of the mounting top plate 3. An inner rotating ring 10 is rotatably mounted in the inner groove. A bottom ring 8 is connected below the inner rotating ring 10. An internal toothed ring 9 is provided on the inner side wall of the bottom ring 8. The internal toothed ring 9 and the external toothed ring 6 are meshed. The upper end of the inner side wall of the inner rotating ring 10 is a sloping structure. An arc-shaped rack 11 is provided on the sloping surface. A first adjusting motor 12 is installed in the mounting top plate 3. The output end of the first adjusting motor 12 is connected to an adjusting gear 13. The adjusting gear 13 and the arc-shaped rack 11 are connected to the inner side wall of the inner rotating ring 10. The rack 11 is engaged; after the chip is fixed and installed by the fixing mechanism, it is necessary to take pictures of the chip. The imaging device 7 is located directly above the chip to take pictures of the chip. The chip is irradiated by the illumination light source 5. Different types of chip defects (surface contamination, internal cracks, circuit abnormalities, etc.) have different reflection, absorption and transmission characteristics of ultraviolet, visible and near-infrared light. Therefore, it is necessary to take pictures under different lighting conditions. When taking pictures, the illumination light source 5 that emits ultraviolet light is used to irradiate the chip vertically downwards. Then, the first adjusting motor 12 is started by controlling the adjusting gear 13 to rotate a certain angle, which causes the inner rotating ring 10 to drive the bottom ring 8 to rotate. Because the inner gear ring 9 and the outer gear ring 6 are engaged, the light source mounting plate 4 can rotate a certain angle, so that the other illumination light sources 5 are turned vertically downwards to irradiate the chip, thus enabling the chip to take pictures under different lighting conditions.

[0034] See Figure 11 The chip defect detector also includes a defect detection system, which includes: a loading and unloading module, a chip fixing module, an image acquisition module, an image preprocessing module, an analysis module, and a data storage and interaction module.

[0035] The loading and unloading module is used for loading and unloading chips. Loading involves accurately transferring the chip from the conveyor belt in the feeding port 2 to the chip tray 16. Unloading involves transferring the inspected chip from the chip tray 16 to the opening of the feeding port 2. The device is equipped with an infrared positioning device and an industrial camera for precise chip positioning. The conveyor belt is precisely controlled by a stepper motor. When loading is completed, a loading completion signal is sent to the chip fixing module to fix the chip.

[0036] The chip fixing module is used to control the fixing mechanism to fix and clamp the chip and adjust its state. After the loading is completed, the chip fixing module adjusts the chip to the image acquisition state and sends an image acquisition signal to the image acquisition module. During unloading, the chip fixing module adjusts the chip's state and moves it to the unloading position, and sends an unloading signal to the unloading module.

[0037] The image acquisition module captures images of the chip through the imaging device 7 and converts the light signal into a 16-bit grayscale digital image. After completing the acquisition of images in all preset bands, a multispectral image set (ultraviolet, visible, and near-infrared images) is formed and transmitted to the image preprocessing module in real time. During image acquisition, the image acquisition module sends a chip adjustment signal to the chip fixing module according to the image acquisition requirements. For example, if the back of the chip needs to be photographed after the front of the chip has been photographed, a signal is sent to the chip fixing module to adjust the chip state. The image acquisition module also includes a light source control unit, which is used to control the switching of the light source, brightness adjustment, and switching between different light sources.

[0038] The image preprocessing module, including an FPGA image accelerator, is used for real-time data processing. It eliminates image noise, corrects positional offsets, and amplifies defect features through algorithms, ensuring the quality of the data input to the analysis module. The core algorithm principle is as follows: Noise suppression: Median filtering is used to preserve the details of defect edges and avoid excessive smoothing; Image registration: Based on the phase correlation method, the coordinates of images in different bands are aligned to ensure that the same defect is in the same position in images of different bands; Contrast enhancement: Adaptive histogram equalization is used to amplify the grayscale difference between defective and normal areas; The workflow of the image preprocessing module is as follows: receiving the multispectral raw image set from the image acquisition module; performing median filtering to denoise the single-band images; registering the ultraviolet and near-infrared images based on the visible band images to eliminate displacement deviations; performing adaptive histogram equalization enhancement on the registered images; outputting the optimized multispectral image set and pixel coordinate mapping table, and transmitting it to the analysis module.

[0039] The analysis module filters suspected defects, extracts features, determines defects, quantifies their size and level, and finally outputs clear defect information from the preprocessed multispectral image set.

[0040] The analysis module's main workflow consists of four steps: Step 1: Screening for potential defects, the core formula is: Where t is the candidate "grayscale boundary line", used to divide image pixels into "bright areas" (which are likely normal areas) and "dark areas" (which are likely suspected defect areas); and These represent the proportions of the bright and dark areas in the entire image, respectively. and These represent the average brightness of the bright and dark areas, respectively. The purpose of this formula is to "find the optimal dividing line t," maximizing the brightness difference between the bright and dark areas. This allows for the rapid separation of suspected defective areas (such as dark spots or lines) from the normal area, initially screening out potentially defective regions and avoiding indiscriminate analysis of the entire image, thus significantly improving subsequent processing efficiency. Each t corresponds to a specific σ²(t): for example, when t=100, σ²(t)=500; when t=120, σ²(t)=800. These two values ​​represent the difference scores corresponding to different dividing lines.

[0041] What we are looking for is the "largest σ²(t)": For example, in the above example, 800, which corresponds to t=120, is a higher score, indicating that when t=120 is used as the dividing line, the difference in brightness between the bright and dark areas is the greatest, and it can most clearly separate suspected defects (such as dark lines and dark spots) from normal areas.

[0042] The second step is to extract features. From the selected suspected areas, extract three types of key information that can distinguish between "defect or normal" and "which type of defect" to form a complete feature package. The first type of key information is grayscale features, which primarily reflect the brightness difference between defects and normal areas. It includes two core formulas: 1. Mean grayscale value: In the formula, M and N are the row and column numbers of the suspected region, respectively. This is the brightness value of each pixel within the region; the entire formula calculates the average brightness of the suspected area. The average brightness of defective areas and normal areas is usually different; for example, cracked areas are darker, and their corresponding average brightness will be smaller. 2. Gray-scale variance: This formula calculates the "uniformity" of brightness within a suspected area; a larger variance indicates a more significant difference in brightness within the area. Defective areas (such as surface contamination) typically have more chaotic brightness, and their variance will be much larger than that of normal areas.

[0043] The second category is spectral characteristics, which primarily reflect the differences in the response of the suspected region under different wavelengths of light. The core formula is the spectral difference degree: In the formula , , These are the brightness values ​​of the suspected area under ultraviolet, visible, and near-infrared light, respectively. , , This is the average brightness of the normal area in the corresponding wavelength band. This formula can calculate the total brightness difference between the suspected area and the normal area in three wavelength bands. Different defects respond differently to different lights (for example, internal defects only show significant differences under near-infrared light). This value can be used to preliminarily distinguish the defect type.

[0044] The third category is morphological characteristics, which primarily reflects the shape features of the defect and includes two core formulas: 1. Roundness: In the formula, A is the area of ​​the defect, and L is the perimeter of the defect, used to calculate the "roundness" of the defect. If C is close to 1, the defect is circular (such as surface contamination spots); if C is less than 0.5, the defect is elongated (such as microcracks). Aspect ratio: This formula calculates the ratio of the length to the width of the circumscribed rectangle of the defect, further confirming the defect shape. For example, the length-to-width ratio of a slender defect (crack) will be very large, while the length-to-width ratio of a circular defect (contamination) will be close to 1.

[0045] The third step is to identify defects. The feature package extracted in the second step is compared with the pre-trained model (which includes known normal region features and various defect features). First, it is determined whether the defect is a defect, and then it is determined what kind of defect it is.

[0046] Core Formula 1 (SVM Decision Function: Determines whether it is a defect): In the formula It is a "feature packet" of the suspected area. It is the "feature package" of the training samples (known as defective or normal regions), and the entire formula is essentially a "comparative similarity". If the calculation result is +1, it is judged as a defect; if it is -1, it is judged as a normal region.

[0047] Core Formula 2 (Softmax function: determines the type of defect): In the formula, C represents the total number of preset defect types (e.g., 3 types: cracks, contamination, and encapsulation defects). This is the score for matching the k-th type of defect in the suspected region. This formula can calculate the probability that the suspected region belongs to each type of defect. If the probability of a certain type of defect is ≥80%, it is determined to be a defect of that type.

[0048] Step 4: Quantify size and level, the core formula is: This indicates that each pixel of the camera corresponds to a 1-micron length on the chip surface. By statistically analyzing the pixel length, width, or area of ​​defects in the image, the actual size can be directly calculated. A simple example of a grading standard is: classifying defects according to their type and actual size, such as minor defects (cracks <50μm, contamination area <100μm²), moderate defects (cracks 50-200μm, contamination area 100-500μm²), and severe defects (cracks >200μm, contamination area >500μm²).

[0049] The final output defect information includes: defect location (chip surface coordinates), whether it is a defect, defect type (crack / contamination / packaging defect, etc.), actual dimensions (length / width / area), and defect severity (minor / moderate / severe). The data storage and interaction module is used to display the inspection progress, defect images, and inspection reports in real time, supports report export, and triggers audible and visual alarms for serious defects.

[0050] This invention discloses a chip defect detector based on multispectral imaging, integrating the detector body, a high-precision fixing mechanism, a multispectral imaging mechanism, and an intelligent defect detection system. The fixing mechanism uses a modular design to achieve stable chip clamping and multi-position adjustment, completely eliminating blind spots. The imaging mechanism, through a multispectral light source surround layout and adaptive angle adjustment, accurately captures the optical response characteristics of different defects. The system, with the help of image preprocessing and intelligent algorithms, completes the accurate identification, quantification, and classification of defects. The equipment effectively solves the pain points of traditional detection methods, such as incomplete coverage, insufficient accuracy, and low efficiency, while possessing the core advantages of automation, omnidirectional coverage, and high precision. It can be widely applied to quality inspection throughout the entire chip manufacturing process, providing key technical support for the high-quality development of the chip industry, and has significant practical value and broad prospects for promotion.

[0051] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A chip defect detector based on multispectral imaging, comprising a detector body, characterized in that: The detector body has a feeding port at the center of the front side and an adjustable fixing mechanism inside. The fixing mechanism is used to fix the chip to be tested. An image acquisition mechanism is set above the fixing mechanism. The image acquisition mechanism is used to acquire images of the chip to be tested under different light irradiation conditions.

2. The chip defect detector based on multispectral imaging according to claim 1, characterized in that: The discharge port is a square channel, with a sealing cover hinged at the open end of the square channel, and a conveyor belt installed inside the square channel.

3. The chip defect detector based on multispectral imaging according to claim 1, characterized in that: The fixing mechanism includes a mounting base, which is located in the middle of the bottom surface of the detector body. A first telescopic rod is installed inside the mounting base, a chip tray is provided on the top of the first telescopic rod, and clamping components are provided on both sides of the chip tray.

4. The chip defect detector based on multispectral imaging according to claim 3, characterized in that: The first telescopic rod and the discharge port are aligned.

5. The chip defect detector based on multispectral imaging according to claim 3, characterized in that: The clamping assembly includes a clamping block, one end of which is fixedly connected to a second telescopic rod. The second telescopic rod is rotatably mounted on a rotating mounting block, which is located on the inner side wall of the main body of the detector. A third adjusting motor is installed inside the rotating mounting block to control the rotation of the second telescopic rod. The other end of the clamping block has a notch, and a square groove is formed on the inner top surface of the notch. A lifting clamping plate is movably disposed in the square groove. A clamping push rod is installed on the top of the clamping block, one end of which passes through the upper part of the clamping block and is connected to the lifting clamping plate. A drive wheel is embedded in the inner bottom surface of the notch.

6. The chip defect detector based on multispectral imaging according to claim 5, characterized in that: Two ball bearings are embedded in the bottom surface of the lifting clamp. A rotating shaft is set at the center of the drive wheel. The rotating shaft is rotatably installed in the clamping block. A first connecting wheel is set on one end of the rotating shaft. The lower part of the clamping block is a hollow structure. A second adjusting motor is set inside the lower part of the clamping block. The output end of the second adjusting motor is connected to the second connecting wheel. The second connecting wheel and the first connecting wheel are connected by a synchronous belt.

7. The chip defect detector based on multispectral imaging according to claim 1, characterized in that: The imaging mechanism includes a mounting top plate, which is located on the inner top surface of the detector body. An imaging device is located at the center of the bottom surface of the mounting top plate, and several sets of adjustable light source components are arranged around the outer side of the imaging device.

8. The chip defect detector based on multispectral imaging according to claim 7, characterized in that: The light source assembly includes a light source mounting plate, a cylindrical mounting block on the bottom surface of a mounting top plate, the light source mounting plate being rotatably mounted on the cylindrical block via an inclined metal rod, an illumination light source mounted on the bottom surface of the light source mounting plate, an external toothed ring on the side wall of the light source mounting plate, an inner groove on the bottom surface of the mounting top plate, an inner rotating ring rotatably mounted in the inner groove, a bottom ring connected below the inner rotating ring, an internal toothed ring on the inner side wall of the bottom ring, the internal toothed ring and the external toothed ring meshing with each other, the upper end of the inner side wall of the inner rotating ring having an inclined structure, an arc-shaped rack on the inclined surface, a first adjusting motor mounted in the mounting top plate, an adjusting gear connected to the output end of the first adjusting motor, the adjusting gear and the arc-shaped rack meshing with each other.

9. The chip defect detector based on multispectral imaging according to claim 8, characterized in that: There are three illumination sources, which respectively illuminate ultraviolet light, visible light and near-infrared light.

10. The chip defect detector based on multispectral imaging according to claim 1, characterized in that: The chip defect detector further includes a defect detection system, which includes: The loading and unloading module is used to load and unload chips. Loading is the accurate transfer of chips from the conveyor belt in the unloading port to the chip tray, and unloading is the transfer of the tested chips from the chip tray to the opening of the unloading port. The chip fixing module is used to control the fixing mechanism to fix and clamp the chip and adjust its state. The image acquisition module captures images of the chip using an imaging device and converts the light signal into a 16-bit grayscale digital image. After acquiring images of all preset wavelengths, it forms a multispectral image set, which is transmitted to the image preprocessing module in real time. The image acquisition module also includes a light source control unit, which controls the switching on and off of the light source, brightness adjustment, and switching between different light sources. The image preprocessing module, including an FPGA image accelerator, is used for real-time data processing. It eliminates image noise, corrects positional offsets, and amplifies defect features through algorithms, ensuring the quality of the data input to the analysis module. The core algorithm principle is as follows: Noise suppression: Median filtering is used to preserve the details of defect edges and avoid excessive smoothing; Image registration: Based on the phase correlation method, the coordinates of images in different bands are aligned to ensure that the same defect is in the same position in images of different bands; Contrast enhancement: Adaptive histogram equalization is used to amplify the grayscale difference between defective and normal areas; The analysis module filters suspected defects, extracts features, judges defects, quantifies their size and level, and finally outputs clear defect information from the preprocessed multispectral image set. The data storage and interaction module is used to display the inspection progress, defect images, and inspection reports in real time, supports report export, and triggers audible and visual alarms for serious defects.

Images