AOI automatic optical detector calibration method and system

Abstract

The invention relates to the technical field of optical detection, in particular to an AOI automatic optical detector calibration method and system, and the method comprises the following steps: obtaining a standard color palette RGB reflection reference, collecting a measured value, calculating an offset contrast, extracting image brightness, generating a correction ratio, converting gray level, extracting abrupt change, screening anomalies, and analyzing a brightness trend to generate a finishing record. And formulating a rule matching correction output comparison table. According to the method, the accuracy and reliability of image detection are enhanced through detailed analysis and accurate correction of color offset and brightness difference, the scheme combines real-time light source data and image information, fine changes are effectively identified, the misjudgment rate is reduced, the accuracy of gray and color correction is improved through comprehensive analysis of light intensity offset and regional brightness trend, and the accuracy of image detection is improved. The detection efficiency and quality of a high-density circuit are ensured, the detailed calibration mechanism remarkably reduces detection errors especially in a high-precision environment, and product quality control and the stability of a production process are optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of optical detection, and particularly to a calibration method and system for an AOI automatic optical inspection instrument. Background Art

[0002] The technical field of optical detection includes technologies for analyzing and judging whether an object to be detected is qualified by using optical devices. This field is widely applied in multiple industries, including the fields of integrated circuits, electronic components, PCBs, etc. Optical detection technology usually illuminates and images the surface of an object by using a camera, sensors, and optical elements, captures reflected light intensity data, and compares it with a standard feature image to determine whether there are defects in the object. With the improvement of industrial manufacturing precision, especially in the production process of integrated circuits and components, optical detection technology has gradually replaced traditional manual detection and has become one of the key technologies for improving production efficiency and quality.

[0003] Among them, the calibration method for an AOI automatic optical inspection instrument refers to a calibration method for an AOI automatic optical inspection instrument system. In the actual detection operation of the prior art, only periodic standard calibration is performed for pixel resolution, dimension indication value, overall color recognition, gray level, and brightness uniformity. The calibration of each parameter is carried out independently, lacking in-depth analysis of the fine changes in channel response and regional brightness differences, and ignoring the local differences between the actual scene in the detection image and the standard calibration environment, making it difficult for the detection device to dynamically adjust the color, gray level, and brightness deviations in the local area; especially in the process of detecting high-precision components such as high-density and fine-line integrated circuits, it is easy to cause the gray level mutation in the solder joint and path intersection areas to be difficult to be effectively distinguished due to subtle color and brightness differences, resulting in local misjudgment and missed detection phenomena, thereby reducing the accuracy and consistency of product detection, making the overall quality control face a high risk of false detection, and seriously affecting the stability of the detection result and production efficiency. Summary of the Invention

[0004] The purpose of the present invention is to solve the disadvantages existing in the prior art and to propose a calibration method and system for an AOI automatic optical inspection instrument.

[0005] In order to achieve the above purpose, the present invention adopts the following technical scheme: A calibration method for an AOI automatic optical inspection instrument, including the following steps: S1: Obtain the reference values of the reflected light intensity of a standard color plate in the three bands of red, green, and blue, collect the reflected light intensity data of the spectral sensor in the key area, calculate the difference between the measured value of the band and the reference value and perform an averaging process to generate a set of reflection offset comparison values; S2: Based on the set of reflection offset comparison values, extract the brightness values of the three channels in the detection image, compare the measured brightness with the corresponding comparison value, calculate the overall change amplitude, and generate a channel response correction ratio; S3: Based on the channel response correction ratio, perform grayscale conversion on the calibrated image, extract the grayscale distribution of the intersection area of the solder joints and the paths, separate the sections with abrupt brightness changes, calculate the difference from the overall brightness, screen the areas beyond the judgment range, and generate a list of grayscale change areas; S4: Based on the list of image grayscale change areas, extract the illumination wavelength and brightness data, compare the regional grayscale with the light source brightness, summarize the brightness change trend and record the average amplitude, and generate a brightness trimming record table; S5: Based on the brightness trimming record table, formulate the color and grayscale comparison rules, apply them to the calibration control of the detector, perform image area matching and correction operations, and generate a calibration comparison table for the detection image.

[0006] As a further solution of the present invention, the reflection offset comparison value group includes the red band deviation, the green band deviation, and the blue band deviation. The channel response correction ratio includes the color change amplitude, the channel brightness correction coefficient, and the overall response adjustment value. The grayscale change area list includes the solder joint abnormal area, the path intersection abnormal area, and the brightness abrupt change abnormal area. The brightness trimming record table includes the illumination wavelength change trend, the regional average brightness deviation, and the light source brightness comparison data. The detection image calibration execution comparison table includes the color matching rule, the grayscale correction rule, and the regional data calibration rule.

[0007] As a further solution of the present invention, the specific steps of S1 are: S101: Obtain the reflection light intensity reference values of the standard color plate in the red, green, and blue bands. Respectively collect the reflection light intensity values of the standard color plate under the illumination conditions of the corresponding light sources in the three bands, distinguish the light intensity data of the bands by channels, establish a three-channel reference data set, and generate a three-channel reflection reference value; S102: Based on the three-channel reflection reference value, collect the reflection light intensity data of the spectral sensor in the integrated circuit solder pad layer, the metal interconnection layer, and the path port area. Respectively perform subtraction processing on the measured values of the bands and the corresponding reference values, obtain the light intensity differences of each sampling point in the red, green, and blue bands, and obtain a three-channel reflection difference set; S103: Call the three-channel reflection difference set, respectively perform averaging processing on the differences of all sampling points in the red, green, and blue bands, combine the three-channel average results as the overall reflection offset index, and obtain the reflection offset comparison value group.

[0008] As a further solution of the present invention, the specific steps of S2 are: S201: Based on the reflection offset comparison value group, obtain the channel brightness values in the detection image, extract the channel brightness of the corresponding pixels in the chip area, pair them with the data of the same band in the reflection offset comparison value group, establish the corresponding relationship between the channel brightness and the offset value, and generate a channel brightness acquisition data group; S202: Based on the channel brightness acquisition data group, perform difference calculation between the brightness of each channel and the corresponding offset value, summarize the channel difference data and calculate the average change value, obtain the overall brightness difference of the channel, and obtain the brightness change amplitude values ​​of the three channels; S203: Calculate the proportion of the channel brightness change to the original brightness according to the brightness change amplitude values ​​of the three channels, and use the uniform conversion to construct a basis for channel brightness adjustment to generate a channel response correction ratio.

[0009] As a further solution of the present invention, the specific calculation formula for calculating the ratio of the channel brightness change to the original brightness is: ; in, Represents the ratio of channel brightness change to original brightness. Representative The brightness value of each channel, Representative The brightness value of the reference channel, Represents the total number of channel brightness data points, Represents the maximum value of the channel brightness.

[0010] As a further solution of the present invention, the specific steps of S3 are: S301: Obtain the channel response correction ratio, perform grayscale conversion on the correction image, extract the image pixel values ​​of the solder joint area and the path intersection area, convert the three-channel brightness values ​​into corresponding grayscale values ​​according to the grayscale conversion formula in the image, and group all grayscale values ​​in the solder joint area and the intersection area into an independent grayscale data set according to the image coordinate position, and generate a regional grayscale distribution set; S302: Based on the regional grayscale distribution set, the image is divided into continuous grayscale segments according to the column order and the row order, and the difference between the grayscale average value of each segment and the average value of the adjacent segment is compared to detect the continuous segment greater than the grayscale mutation threshold, and the adjacent mutation segments are merged according to the threshold range to generate the brightness mutation segment interval; S303: Call the brightness mutation segment interval, calculate the difference between the grayscale value of each mutation segment and the average grayscale value of the entire image, filter the segments whose difference is greater than the grayscale judgment threshold, and output the image area index group with significant grayscale difference according to the screening result to generate a grayscale change area list.

[0011] As a further solution of the present invention, the specific steps of S4 are: S401: Obtaining the image acquisition timestamp of each area in the grayscale change area list, acquiring the light source wavelength data and brightness data corresponding to the acquisition time point, and grouping the wavelength values ​​and brightness values ​​of the same area at the different time points in chronological order to form a combination of illumination data for each grayscale abnormal area, and generate a regional illumination information combination value; S402: Based on the regional illumination information combination value, extract the grayscale value of each region and the light source brightness value at the corresponding time point, calculate the difference between the two, and classify and arrange them by wavelength type, obtain the contrast relationship between grayscale and brightness changes under different wavelengths, and generate a regional grayscale brightness change relationship group; S403: According to the regional grayscale brightness change relationship group, the grayscale and brightness differences in each wavelength group are summarized, the average amplitude of the differences in the same group is calculated, all the grouped average values ​​are correspondingly integrated with the regional information, and imported into a unified data form structure to generate a brightness adjustment record table.

[0012] As a further solution of the present invention, the specific calculation formula of the light source wavelength data and brightness data corresponding to the acquisition time point is: ; Calculate the lighting combination and generate the regional lighting information combination value; in, Represents at a point in time and between, area and Region The combined value of lighting data, Represents at a point in time and The corresponding collection The difference in wavelength data, and Represents the time points and The corresponding collection Brightness data, Representative The time interval corresponding to the time period, Represents the number of wavelength data points involved in the calculation, and the summation symbol Represents the cumulative calculation of all wavelength data points.

[0013] As a further solution of the present invention, the specific steps of S5 are: S501: extracting the channel response value and the brightness trimming value of the image area based on the brightness trimming record table and the channel response correction ratio, performing combination processing and classifying according to the channel matching rule, establishing the color and brightness relationship corresponding to the area, and generating a color grayscale matching relationship group; S502: Call the color-grayscale matching relationship group and the grayscale change area list, extract the channel brightness and grayscale mutation range of the area to be calibrated, compare according to the corresponding relationship, screen the areas that meet the conditions, and obtain the area channel comparison result group; S503: According to the area channel comparison result group and the brightness adjustment value in the brightness trimming record table, match the color and grayscale combination of the area to be calibrated, determine the correction number and replace the corresponding data, and obtain the detection image calibration execution comparison table.

[0014] AOI automatic optical inspection instrument calibration system, including: The reflection acquisition and analysis module obtains the reflection light intensity reference values of the standard color plate in the red, green, and blue bands, collects the reflection light intensity data of the spectral sensor at the pad layer, metal interconnection layer, and via port, calculates the difference between the measured value and the reference value, averages the difference of each band, integrates the band average offset, and generates a reflection offset comparison value group; The channel brightness correction module extracts the red, green, and blue channel brightness values in the image based on the reflection offset comparison value group, compares the measured brightness with the comparison value, calculates the channel brightness difference and the overall change range, converts it into a ratio form, and generates a channel response correction ratio; The image grayscale recognition module performs grayscale conversion on the image according to the channel response correction ratio, extracts the grayscale distribution at the intersection of the solder joints and the vias, compares the sectional grayscale with the overall average value, determines whether the grayscale difference of the mutation section exceeds the range, screens the abnormal sections, and generates a grayscale change area list; The brightness trend extraction module calls the grayscale change area list, collects the illumination wavelength and light source brightness of the abnormal area at the corresponding time period, compares the grayscale value with the light source brightness, classifies the change trend, calculates the average change range, and generates a brightness trimming record table; The area matching and calibration module calls the brightness trimming record table, the grayscale change area list, and the channel response correction ratio, extracts the channel brightness and grayscale values of the area to be calibrated, determines the correction rule index according to the correction ratio and the adjustment record, performs area comparison and adjustment, and generates a detection image calibration execution comparison table.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, by analyzing and accurately correcting color offset and brightness difference in detail, the accuracy and reliability of image detection are enhanced. This solution combines real-time light source data and image information, effectively identifies subtle changes and reduces the false judgment rate. By comprehensively analyzing light intensity offset and regional brightness trend, the accuracy of grayscale and color correction is improved, ensuring the efficiency and quality of high-density circuit detection. This meticulous calibration mechanism significantly reduces detection errors especially in a high-precision environment, optimizing product quality control and the stability of the production process. Description of the Drawings

[0016] Figure 1 Schematic diagram of the step flow of the present invention Figure 2 System module diagram of the present invention. Specific implementation manners

[0017] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0018] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation to the present invention. In addition, in the description of the present invention, the meaning of "a plurality of" is two or more unless otherwise specifically defined.

[0019] Please refer to Figure 1 , the calibration method of the AOI automatic optical inspection instrument, including the following steps: S1: Obtain the reference values of the reflected light intensities of the standard color plate in the three bands of red, green, and blue, collect the reflected light intensity data of the spectral sensor in the key area of the integrated circuit, calculate the difference between the measured value and the reference value of each band, and perform an average process on the band error to generate a set of reflection offset comparison values; S2: Based on the set of reflection offset comparison values, extract the brightness values of the three channels in the detection image, compare the measured brightness with the corresponding comparison value, calculate the overall change amplitude, and generate a channel response correction ratio; S3: Based on the channel response correction ratio, perform gray conversion on the corrected image, extract the gray distribution of the intersection area of the solder joints and the paths, separate the sections with sudden brightness changes, calculate the difference between the section with sudden brightness change and the overall brightness, screen the areas exceeding the judgment range, and generate a list of gray change areas; S4: Based on the list of gray change areas, extract the illumination wavelength and brightness data during acquisition, compare the gray level of each area with the light source brightness at the corresponding time point, classify the change trends, record the average change amplitude, and generate a brightness trimming record table; S5: Based on the brightness trimming record table, combine the channel response correction ratio and the list of image gray change areas, establish a color and gray comparison rule, apply it to the calibration control of the inspection instrument, perform area matching and data correction on the image, and generate a calibration execution comparison table for the detection image.

[0020] The reflection offset reference value group includes the red band deviation, the green band deviation, and the blue band deviation. The channel response correction ratio includes the color change amplitude, the channel brightness correction coefficient, and the overall response adjustment value. The grayscale change area list includes the solder joint anomaly area, the path intersection anomaly area, and the brightness mutation anomaly area. The brightness trimming record table includes the light wavelength change trend, the regional average brightness deviation, and the light source brightness comparison data. The detection image calibration execution comparison table includes the color matching rule, the grayscale correction rule, and the regional data calibration rule.

[0021] The specific steps of S1 are as follows: S101: Obtain the reference reflection light intensity values of the standard color plate in the red, green, and blue bands. Respectively collect the reflection light intensity values of the standard color plate under the illumination conditions of the corresponding light sources in the three bands. Differentiate the light intensity data of the bands by channels, establish a three-channel reference data set, and generate a three-channel reflection reference value. To obtain the reference reflection light intensity values of the standard color plate in the red, green, and blue bands, a standard color plate with a uniform surface color and stable reflection performance needs to be selected as the collection object. Combined with the commonly used light sources for integrated circuit detection, it is irradiated in different bands in the experimental environment. The light source wavelengths are respectively set to 625 nm, 530 nm, and 470 nm, corresponding to the red, green, and blue channels respectively. Under each band, a spectral sensor is used to record the reflection light intensity data at multiple positions of the standard color plate, ensuring that the illumination angle is 90 degrees perpendicular to avoid interference introduced by the change of the incident angle. During the collection, the light source brightness needs to be controlled within 1000 lux. Continuously collect data at multiple points in the same band through the sensor to ensure representativeness. For example, 5 points are collected in the red channel, and the reflection values are 2.1, 2.0, 2.2, 2.1, 2.1 mW / cm². Calculate the average value of this group of data as the reference light intensity value of the red channel. The same process is repeated for the other two channels. The collected values are classified according to channel attribution, and after marking the channel information, a three-channel data set is constructed. Each group of data contains the band identifier, the sampling point position, and the reflection value. The average value of the reflection data in each channel is used as the reference reflection light intensity of the channel under the standard color plate, and is respectively named the red channel reference value, the green channel reference value, and the blue channel reference value, forming a three-channel reflection reference value.

[0022] S102: Based on the three-channel reflection reference value, collect the reflection light intensity data of the spectral sensor in the integrated circuit pad layer, metal interconnection layer, and path port area. Respectively perform subtraction processing on the measured values of the bands and the corresponding reference values to obtain the light intensity differences of each sampling point in the red, green, and blue bands, and obtain a three-channel reflection difference set. Based on the three-channel reflection reference value, in the process of integrated circuit image acquisition, three areas, namely the pad layer, the metal interconnection layer and the via port, are selected as detection targets. An equal number of sampling points are arranged in these areas respectively, and the lighting conditions are kept consistent. The reflected light intensity data of each point is collected in turn under the illumination of three light sources. The sensor records the actual reflection value under the current light source and compares it with the three-channel reference value established in the early stage. The difference between the actual reflection and the standard value of each point under the red, green and blue channels is calculated, and each difference and its corresponding channel and point information are recorded to form a complete sampling point reflection difference detailed data. All difference data are classified and grouped by channel, and integrated into a data set containing spatial position, channel information and difference content, namely the three-channel reflection difference set.

[0023] S103: calling a three-channel reflection difference set, averaging the differences of all sampling points in the red, green and blue bands respectively, combining the three-channel average results as an overall reflection offset index, and obtaining a reflection offset reference value group; The three-channel reflection difference set is called to aggregate the reflection differences of all points collected in each band. The data of the red, green and blue bands are summarized by channel respectively, and the average difference level of all sampling points in each channel is counted as the overall offset performance of the channel in the target area. For example, the red channel counts the reflection differences of 10 points to get the average difference value, and the same statistical process is repeated for the green and blue channels. The average differences of the three channels are combined to form a complete overall offset expression structure of reflected light intensity. This combined data can be used as the basis for subsequent color correction and channel brightness trimming. The results are uniformly named as reflection offset control value group.

[0024] The specific steps of S2 are: S201: Based on the reflection offset reference value group, the channel brightness value in the detection image is obtained, the channel brightness of the corresponding pixel in the chip area is extracted, and the channel brightness is matched with the same band data in the reflection offset reference value group, and the corresponding relationship between the channel brightness and the offset value is established to generate a channel brightness acquisition data group; Based on the reflection offset reference value group, it is first necessary to extract the brightness value of each channel in the detected image and determine the pixel data of the target chip area. This process extracts the RGB channel brightness values (or brightness data in other color spaces) of the image through an image processing tool and selects relevant pixels within the chip area for analysis. For example, a certain area is selected in the detected image and its brightness data is obtained. Suppose the RGB brightness values of a certain pixel are 255, 240, and 230. Next, these values are compared with the data of the same band in the reflection offset reference value group. During the comparison process, the brightness data of each band is matched one by one with the offset data in the reflection offset reference value group, so that the corresponding relationship between the channel brightness and the offset value can be established through this group of data. The key to this process lies in how to accurately extract the brightness of each channel in the image, accurately perform the comparison and matching, and generate a channel brightness acquisition data group for subsequent calculations.

[0025] S202: Based on the channel brightness acquisition data group, calculate the difference between the brightness of each channel and the corresponding offset value, summarize the channel difference data and calculate the average change value to obtain the overall channel brightness difference, and obtain the brightness change amplitude values of the three channels; After obtaining the channel brightness acquisition data group, the next task is to calculate the difference between the brightness of each channel and the corresponding offset value. This difference calculation can be obtained by comparing the brightness data of each channel with its corresponding offset value. For example, the brightness data of the red channel may be several groups of values, while the corresponding offset value is another group of data. For each group of brightness and offset values, calculate the difference between them. By summarizing all the difference data, an average change amount of the brightness of each channel can be obtained. Suppose the difference calculated for the red channel is within a certain fixed range, and these differences are finally summarized into an average change value, which represents the degree of brightness change of the red channel in the entire acquisition data. A similar calculation method is applicable to the brightness difference calculation of the green and blue channels, obtaining the brightness change amplitude of each channel, and combining the calculation results of all channels to further obtain the overall brightness difference value of the image.

[0026] S203: According to the brightness change amplitude values of the three channels, calculate the proportion of the channel brightness change to the original brightness, uniformly convert it to construct the basis for channel brightness adjustment, and generate the channel response correction ratio; The specific calculation formula for calculating the proportion of the channel brightness change to the original brightness is: ; Among them, represents the proportion of the channel brightness change to the original brightness, represents the th channel brightness value, represents the The brightness value of a reference channel, represents the total number of channel brightness data points, represents the maximum value of the channel brightness; In this formula, represents the proportion of the channel brightness change to the original brightness, aiming to evaluate the relative effect of the channel brightness correction. The numerical values of all parameters are obtained through the data monitoring or acquisition process and are associated with the performance indicators of the actual device.

[0027] represents the actual brightness value of the th channel, and these values are obtained through a brightness sensor or an image processing device. In actual operation, the brightness value may be obtained by the sensor reading the image data and performing numerical processing. For example, it is collected through the image brightness distribution under the lighting device, and the numerical range is from 0 to 255.

[0028] is obtained by comparing a set of standard or original reference images. For example, assume that the channel brightness of the standard image used for comparison is .

[0029] represents the amount of channel data for comparison. In this example, assume that there are a total of data points, that is, the brightness of five channels is compared.

[0030] This value is determined by the scale of the brightness sensor, and the maximum value is 255, representing the maximum limit of the image brightness.

[0031] Based on the collected data, assume we have the following actual brightness value data (unit: brightness value): - The brightness value of the reference channel: ; First, calculate the sum of the absolute values of the brightness differences: ; ; Then, calculate the sum of the squares of the brightness: ; ; ; ; Calculate the square root part of the denominator: ; Finally, substitute these calculation results: ; The result shows that the ratio of channel brightness change to original brightness is about 0.0255, or 2.55%. This ratio reflects the difference between the actual brightness value and the reference brightness value, and is normalized by the maximum brightness.

[0032] The specific steps of S3 are: S301: Obtain channel response correction ratios, perform grayscale conversion on the correction image, extract image pixel values ​​of the solder joint area and the path intersection area, convert the three-channel brightness values ​​into corresponding grayscale values ​​according to the grayscale conversion formula in the image, and group all grayscale values ​​in the solder joint area and the intersection area into independent grayscale data sets according to the image coordinate positions, and generate a regional grayscale distribution set; In the actual application of image processing, the channel response correction ratio of the image is first obtained, and the image is gray-scale converted by using the ratio. Specifically, for each pixel, the three-channel (red, green, and blue) values ​​of the image need to be converted into a single gray-scale value through the standard weighted gray-scale conversion formula. For example, if the red, green, and blue channel values ​​of a pixel in the image are 120, 200, and 150, respectively, then the corresponding gray-scale value can be obtained by calculation. This processing can effectively unify the brightness information of the image and lay the foundation for subsequent processing steps. Next, by extracting the pixel values ​​of the solder joint area and the intersection area of ​​the via, and grouping their gray-scale values, a data set containing the gray-scale distribution of these areas can be generated. For example, if the gray-scale value of the solder joint area is relatively uniform and prominent, then its gray-scale data will be concentrated within a certain range, which is convenient for subsequent analysis. In this way, the gray-scale information of the image is accurately extracted, providing the necessary basic data support for subsequent image processing.

[0033] S302: Based on the regional grayscale distribution set, the image is divided into continuous grayscale segments according to the column and row order, and the difference between the grayscale average value of each segment and the average value of the adjacent segment is compared to detect the continuous segments greater than the grayscale mutation threshold, and the adjacent mutation segments are merged according to the threshold range to generate the brightness mutation segment interval; Based on the regional grayscale distribution set, each area of ​​the image will be divided into several continuous grayscale segments. At this time, the average grayscale value of each segment needs to be calculated first. For example, if the grayscale value of a column of pixels in the image is a series of data, the average grayscale value of the column can be obtained by calculating the average of these data. After calculating the average value of each area, the system will compare the average values ​​of adjacent grayscale segments. If the difference exceeds the set grayscale mutation threshold, it is considered that there is a significant brightness mutation area. At this time, if it is found that the brightness change between some grayscale segments is too large (for example, the change in grayscale value exceeds 10), the system will automatically mark these areas as brightness mutations. Next, if the difference between multiple adjacent mutation segments is small, the boundaries will be merged to merge them into an overall area, which helps to identify more significant brightness mutation areas.

[0034] S303: calling the brightness mutation segment interval, calculating the difference between the gray value of each mutation segment and the average gray value of the entire image, screening the segments whose difference is greater than the gray judgment threshold, and outputting the image region index group with significant gray difference according to the screening result, and generating a gray change region list; After obtaining the brightness mutation segment, the system will further process the grayscale value of each mutation segment and calculate the difference with the average grayscale value of the entire image. If the difference exceeds the set grayscale judgment threshold (for example, the threshold is set to 5), the mutation segment is considered to have a significant grayscale difference and requires further analysis. Through this process, the system can filter out those areas that are significantly different from the entire image in grayscale. For example, if the grayscale value of a mutation area is 135 and the average grayscale value of the entire image is 120, then their difference is 15, which is greater than the set difference threshold of 5, so the area will be marked as an area with significant grayscale changes. Through this judgment process, the system finally generates a list of grayscale change areas to help further analyze and process the image in detail.

[0035] The specific steps of S4 are: S401: Obtain the image acquisition timestamp of each area in the grayscale change area list, acquire the light source wavelength data and brightness data corresponding to the acquisition time point, and aggregate the wavelength values ​​and brightness values ​​of the same area at the different time points in chronological order to form a combination of illumination data for each grayscale abnormal area, and generate a regional illumination information combination value; The specific calculation formula for the light source wavelength data and brightness data corresponding to the acquisition time point is: ; Calculate the lighting combination and generate the regional lighting information combination value; in, Represents at a point in time and Between the regions and the region The combined value of the light intensity data represents the difference in the wavelength data collected at the time points and corresponding to the wavelength data. and respectively represent the and corresponding brightness data collected at the time points. represents the time interval corresponding to the time period, and the summation symbol indicates the cumulative calculation for all wavelength data points; Parameter definition and acquisition method: By comparing the changes in the light source wavelength and brightness data at two time points, the changes in the light intensity characteristics of the region are reflected.

[0036] The calculation method is: where and are the measured values of the and wavelengths at the time points respectively. The wavelength data is obtained by measuring the same region multiple times with a spectrometer at different time points, and the unit is nanometers (nm).

[0037] The calculation method is: ; where and are the and time stamps at the

[0038] and time points respectively, and the unit is seconds (s). The time stamps are obtained by recording with a high-precision clock during the image acquisition process.

[0039] Determined according to the experimental design and the required spectral resolution, and the value range is generally between 1 and 100.

[0040] Formula calculation and derivation process: Assume at the time point and light source data at three wavelengths ( ) were collected. The specific steps are as follows: Calculate the wavelength difference : For example, at time points and , the wavelength data collected are respectively: For wavelength , , , then the wavelength difference is: ; For wavelength , , , then the wavelength difference is: ; For wavelength , , , then the wavelength difference is: ; Calculate the time interval : Assume that the timestamps of time points and are respectively: ; ; Then the time interval is calculated as: ; Calculate the sum of squares of the luminance data: Assume that the luminance data collected at time points and are as follows: For wavelength , , ; For wavelength , , ; For wavelength , , ; Then the sum of squares of each wavelength data is calculated as: ; ; ; Calculate the numerator and denominator: The numerator is calculated as: ; The denominator is calculated as: ; Calculate the combined value of the light data : Combined value of the light data Is calculated as: ; Meaning of the calculation result: Represents between time points and the combined value of the light data in the regions and region This value reflects the combined effect of the change in the light source wavelength and the change in brightness between the two time points. A higher combined value indicates that the light characteristics of this region have changed significantly during this period.

[0041] S402: Based on the combined value of the regional light information, extract the gray value of each region and the light source brightness value corresponding to the time point, calculate the difference between the two, and arrange them in categories according to the wavelength type to obtain the contrast relationship between the gray value and the brightness change under different wavelengths, and generate a set of regional gray-brightness change relationships; Based on the extracted combined value of the regional light information, the next step is to extract the gray value of each region and calculate the difference with the light source brightness value corresponding to the time point. At this time, the gray value reflects the brightness change of the regional surface or object, while the light source brightness value records the intensity of the ambient light at this time. For example, at a certain moment, the gray value of a certain region is 150, and the corresponding light source brightness is 2000 Lux. At another time point, the gray value of this region becomes 160, and the light source brightness increases to 2200 Lux. At this time, the change difference between the two can be calculated: the gray difference is 10, and the brightness difference is 200. In this way, the trend of the regional gray value changing with the light can be evaluated under different time points or different light source conditions. Then, these data are arranged in categories according to the light sources of different wavelengths, such as wavelengths of 450 nm, 550 nm, 650 nm, etc. These wavelengths represent different light source conditions. After classification, the relationship between the gray value and the brightness change under different wavelengths can be obtained, which provides basic data for the subsequent contrast analysis of the gray-brightness change and helps to understand the influence of different wavelength light sources on the image gray value.

[0042] S403: Summarize the differences between grayscale and brightness for each wavelength grouping according to the regional grayscale-brightness change relationship group, calculate the average amplitude of the differences within the same group, and integrally correspond the average values after all groupings with the regional information and import them into a unified data form structure to generate a brightness trimming record table; According to the regional grayscale-brightness change relationship group, next, it is necessary to summarize the differences between grayscale and brightness in each wavelength grouping and calculate the average amplitude of the differences within the same group. At this time, first, summarize the difference data in each wavelength group, aggregate the grayscale and brightness difference data into one group, and calculate the average value of these differences. For example, if within a certain wavelength range, the grayscale and brightness difference data of multiple regions are as follows: the grayscale differences are 10, 12, and 9 respectively, and the brightness differences are 200, 180, and 220 respectively, by summarizing these difference data, the average difference value of this wavelength group can be calculated. The final result will show the average situation of the grayscale and brightness differences in different wavelengths in the region. These results will be combined with the regional information to form a complete brightness trimming record table. By sorting out the grayscale and brightness differences in each wavelength group, this table can provide a basis for subsequent light adjustment and further improve the quality control accuracy of the product.

[0043] The specific steps of S5 are as follows: S501: Based on the brightness trimming record table and the channel response correction ratio, extract the channel response value and the brightness trimming value of the image region, perform combined processing and classify according to the channel comparison rule, establish the corresponding color and brightness relationship of the region, and generate a color-grayscale matching relationship group; First, obtain the brightness trimming record table, and the content in the record table includes the brightness trimming parameters for different image regions. The channel response correction ratio reflects the color response characteristics of the image region in different channels. For each region, there is a corresponding channel response correction value and brightness trimming value. Through these correction values, the actual color response situation of the image region can be obtained. Next, classify the image region according to the channel comparison rule, and the channel comparison rule can be distinguished according to the range of the response correction ratio. For example, if the red channel response correction ratio of a certain region is 1.1 and the green channel response correction ratio is 0.9, then this region may belong to a specific category. By the change of these ratios, establish the corresponding color and brightness relationship of the region. Finally, through these relationships, generate a color-grayscale matching relationship group, which can be used in subsequent image calibration to achieve precise brightness adjustment of the target region.

[0044] S502: Call the color-grayscale matching relationship group and the grayscale change region list, extract the channel brightness and the grayscale mutation range of the region to be calibrated, compare according to the corresponding relationship, screen out the regions that meet the conditions, and obtain the regional channel comparison result group; First, extract the color-gray matching relationship group, which includes the relationship model between the predefined channel response and the gray change. The list of gray change regions contains the gray mutation conditions of each region in the image. Gray mutation refers to the sharp change in brightness within a certain region of the image, which occurs at the edges of objects or in texture regions. For each region to be calibrated, the channel brightness and the gray mutation range of this region need to be extracted. For example, there may be a mutation region with a brightness ranging from 80 to 160 in a certain region, which needs to be compared with the corresponding values in the color-gray matching relationship group to determine whether it meets the matching conditions. The comparison process can be determined by simple rules. If the region brightness coincides with the gray range in the matching relationship group, it is considered that this region meets the conditions. According to these comparison results, the regions that meet the requirements are screened out, and finally the region channel comparison result group is obtained. This result group can provide reference data for subsequent color calibration.

[0045] S503: According to the region channel comparison result group and the brightness adjustment values in the brightness trimming record table, match the color and gray combination of the region to be calibrated, determine the correction number and replace the corresponding data to obtain the calibration execution comparison table for the detected image; First, it is necessary to combine the region channel comparison result group and the brightness adjustment values in the brightness trimming record table. The brightness trimming record table provides the brightness correction parameters for different image regions. These parameters include the adjusted brightness value and the corresponding correction coefficient. According to the channel comparison results, the color and gray combination of the region to be calibrated can be matched with the brightness adjustment values to ensure that the brightness correction during calibration meets the expectations. If the color and gray combination of a certain region matches a certain brightness adjustment value, the correction number of this region is the number of this brightness adjustment value. For example, assume that the brightness of the red channel in a certain region is 120, the brightness of the green channel is 130, and the brightness of the blue channel is 110. Through comparison, the most suitable brightness adjustment value is found and corresponding to this region. Finally, the obtained calibration execution comparison table for the detected image contains the correction number of each region and replaces the color and brightness information in the original data, so as to obtain the calibrated image data.

[0046] Please refer to Figure 2 , the calibration system of the AOI automatic optical inspection instrument, includes: The reflection acquisition and analysis module obtains the reflection light intensity reference values of the standard color plate in the red, green, and blue bands, collects the reflection light intensity data of the spectral sensor at the pad layer, the metal interconnection layer, and the via port, calculates the difference between the measured value and the reference value, performs an average process on the difference of each band, integrates the average offset of the band, and generates a reflection offset comparison value group; The channel brightness correction module, based on the reflection offset comparison value group, extracts the brightness values of the red, green, and blue channels in the image, compares the measured brightness with the comparison value, calculates the channel brightness difference and the overall change amplitude, converts it into a ratio form, and generates a channel response correction ratio; The image grayscale recognition module corrects the ratio according to the channel response, performs grayscale conversion on the image, extracts the grayscale distribution at the intersection of the solder joints and the paths, compares the grayscale of the section with the overall average value, determines whether the grayscale difference of the mutation section exceeds the range, screens the abnormal sections, and generates a list of grayscale change regions; The brightness trend extraction module calls the list of grayscale change regions, collects the light wavelength and light source brightness corresponding to the abnormal area during the corresponding period, compares the grayscale value with the light source brightness, classifies the change trend, calculates the average change amplitude, and generates a brightness trimming record table; The region matching and calibration module calls the brightness trimming record table, the list of grayscale change regions and the channel response correction ratio, extracts the channel brightness and grayscale value of the region to be calibrated, determines the correction rule index according to the correction ratio and the adjustment record, performs region comparison and adjustment, and generates a calibration execution comparison table for the detected image.

[0047] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any person skilled in the art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical content of the technical solution of the present invention, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. AOI automatic optical inspection instrument calibration method, characterized in that: The following steps are involved: S1: Obtain the reference values ​​of the reflected light intensity of the standard color plate in the red, green and blue bands, collect the reflected light intensity data of the spectral sensor in the key area, calculate the difference between the actual measured value and the reference value of the band and perform average processing to generate a reflection offset comparison value group; S2: based on the reflection offset reference value group, extracting the brightness values ​​of the three channels in the detection image, comparing the measured brightness with the corresponding reference value, calculating the overall change amplitude, and generating a channel response correction ratio; S3: Based on the channel response correction ratio, grayscale conversion is performed on the corrected image, the grayscale distribution of the intersection area of ​​the solder joint and the via is extracted, the brightness mutation section is separated and the difference with the overall brightness is calculated, the area beyond the judgment range is screened, and a list of grayscale change areas is generated; S4: Based on the image grayscale change area list, extract the light wavelength and brightness data, compare the regional grayscale with the light source brightness, summarize the brightness change trend and record the average amplitude, and generate a brightness adjustment record table; S5: Based on the brightness adjustment record table, formulate color and grayscale comparison rules, apply them to detector calibration control, perform image area matching and correction operations, and generate a detection image calibration comparison table.

2. The AOI automatic optical inspection instrument calibration method according to claim 1, characterized in that: The reflection offset control value group includes red band deviation, green band deviation, and blue band deviation. The channel response correction ratio includes color change amplitude, channel brightness correction coefficient, and overall response adjustment value. The grayscale change area list includes solder joint abnormal area, path intersection abnormal area, and brightness mutation abnormal area. The brightness trimming record table includes light wavelength change trend, regional average brightness deviation, and light source brightness comparison data. The detection image calibration execution control table includes color matching rules, grayscale correction rules, and regional data calibration rules.

3. The AOI automatic optical inspection instrument calibration method according to claim 1, characterized in that: The specific steps of S1 are: S101: Obtaining the reflected light intensity reference values ​​of the standard color plate in the red, green and blue bands, respectively collecting the reflected light intensity values ​​of the standard color plate under the illumination conditions of the three bands corresponding to the light sources, performing channel differentiation on the light intensity data of the bands, establishing a three-channel reference data set, and generating a three-channel reflection reference value; S102: Based on the three-channel reflection reference values, the reflected light intensity data of the spectral sensor at the integrated circuit pad layer, the metal interconnection layer and the via port area are collected, and the actual measured values ​​of the bands and the corresponding reference values ​​are respectively subjected to difference processing to obtain the light intensity difference of each sampling point in the red, green and blue bands, and obtain a three-channel reflection difference value set; S103: calling the three-channel reflection difference set, averaging the differences of all sampling points in the red, green and blue bands respectively, combining the three-channel average results as an overall reflection shift index, and obtaining a reflection shift reference value group.

4. The AOI automatic optical inspection instrument calibration method according to claim 3, characterized in that: The specific steps of S2 are: S201: Based on the reflection offset reference value group, the channel brightness value in the detection image is obtained, the channel brightness of the corresponding pixel in the chip area is extracted, and the channel brightness is paired with the same band data in the reflection offset reference value group, and the corresponding relationship between the channel brightness and the offset value is established to generate a channel brightness acquisition data group; S202: Based on the channel brightness acquisition data group, perform difference calculation between the brightness of each channel and the corresponding offset value, summarize the channel difference data and calculate the average change value, obtain the overall brightness difference of the channel, and obtain the brightness change amplitude values ​​of the three channels; S203: Calculate the proportion of the channel brightness change to the original brightness according to the brightness change amplitude values ​​of the three channels, and use the uniform conversion to construct a basis for channel brightness adjustment to generate a channel response correction ratio.

5. The AOI automatic optical inspection instrument calibration method according to claim 4, characterized in that: The specific calculation formula for the ratio of the channel brightness change to the original brightness is: ； in, Represents the ratio of channel brightness change to original brightness. Representative The brightness value of each channel, Representative The brightness value of the reference channel, Represents the total number of channel brightness data points, Represents the maximum value of the channel brightness.

6. The AOI automatic optical inspection instrument calibration method according to claim 4, characterized in that: The specific steps of S3 are: S301: Obtain the channel response correction ratio, perform grayscale conversion on the correction image, extract the image pixel values ​​of the solder joint area and the path intersection area, convert the three-channel brightness values ​​into corresponding grayscale values ​​according to the grayscale conversion formula in the image, and group all grayscale values ​​in the solder joint area and the intersection area into an independent grayscale data set according to the image coordinate position, and generate a regional grayscale distribution set; S302: Based on the regional grayscale distribution set, the image is divided into continuous grayscale segments according to the column order and the row order, and the difference between the grayscale average value of each segment and the average value of the adjacent segment is compared to detect the continuous segment greater than the grayscale mutation threshold, and the adjacent mutation segments are merged according to the threshold range to generate the brightness mutation segment interval; S303: Call the brightness mutation segment interval, calculate the difference between the grayscale value of each mutation segment and the average grayscale value of the entire image, filter the segments whose difference is greater than the grayscale judgment threshold, and output the image area index group with significant grayscale difference according to the screening result to generate a grayscale change area list.

7. The AOI automatic optical inspection instrument calibration method according to claim 6, characterized in that: The specific steps of S4 are: S401: Obtaining the image acquisition timestamp of each area in the grayscale change area list, acquiring the light source wavelength data and brightness data corresponding to the acquisition time point, and grouping the wavelength values ​​and brightness values ​​of the same area at the different time points in chronological order to form a combination of illumination data for each grayscale abnormal area, and generate a regional illumination information combination value; S402: Based on the regional illumination information combination value, extract the grayscale value of each region and the light source brightness value at the corresponding time point, calculate the difference between the two, and classify and arrange them by wavelength type, obtain the contrast relationship between grayscale and brightness changes under different wavelengths, and generate a regional grayscale brightness change relationship group; S403: According to the regional grayscale brightness change relationship group, the grayscale and brightness differences in each wavelength group are summarized, the average amplitude of the differences in the same group is calculated, all the grouped average values ​​are correspondingly integrated with the regional information, and imported into a unified data form structure to generate a brightness adjustment record table.

8. The AOI automatic optical inspection instrument calibration method according to claim 7, characterized in that: The specific calculation formula for the light source wavelength data and brightness data corresponding to the acquisition time point is: ； Calculate the lighting combination and generate the regional lighting information combination value; in, Represents at a point in time and between, area and Region The combined value of lighting data, Represents at a point in time and The corresponding collection The difference in wavelength data, and Represents the time points and The corresponding collection Brightness data, Representative The time interval corresponding to the time period, Represents the number of wavelength data points involved in the calculation, and the summation symbol Represents the cumulative calculation of all wavelength data points.

9. The AOI automatic optical inspection instrument calibration method according to claim 7, characterized in that: The specific steps of S5 are: S501: extracting the channel response value and the brightness trimming value of the image area based on the brightness trimming record table and the channel response correction ratio, performing combination processing and classifying according to the channel matching rule, establishing the color and brightness relationship corresponding to the area, and generating a color grayscale matching relationship group; S502: calling the color grayscale matching relationship group and the grayscale change area list, extracting the channel brightness and grayscale mutation range of the area to be calibrated, performing comparison according to the corresponding relationship, screening the area meeting the conditions, and obtaining the area channel comparison result group; S503: Match the color and grayscale combination of the area to be calibrated according to the area channel comparison result group and the brightness adjustment value in the brightness modification record table, determine the correction number and replace the corresponding data, and obtain the detection image calibration execution comparison table.

10. AOI automatic optical inspection instrument calibration system, characterized in that: According to the AOI automatic optical inspection instrument calibration method according to any one of claims 1 to 9, the system comprises: The reflection acquisition and analysis module obtains the reflection light intensity reference values ​​of the standard color plate in the red, green and blue bands, collects the reflection light intensity data of the spectral sensor at the pad layer, metal interconnect layer and via port, calculates the difference between the measured value and the reference value, averages the difference in each band, integrates the band average offset, and generates a reflection offset comparison value group; The channel brightness correction module extracts the red, green and blue channel brightness values ​​in the image based on the reflection offset reference value group, compares the measured brightness with the reference value, calculates the channel brightness difference and finds the overall change amplitude, converts it into a ratio form, and generates a channel response correction ratio; The image grayscale recognition module performs grayscale conversion on the image according to the channel response correction ratio, extracts the grayscale distribution at the intersection of the solder joint and the path, compares the grayscale of the segment with the overall mean, determines whether the grayscale difference of the mutation segment exceeds the range, screens the abnormal segment, and generates a list of grayscale change areas; The brightness trend extraction module calls the grayscale change area list, collects the light wavelength and light source brightness of the abnormal area during the corresponding period, compares the grayscale value with the light source brightness, classifies the change trend, calculates the average change amplitude, and generates a brightness adjustment record table; The regional matching calibration module calls the brightness adjustment record table, the grayscale change area list and the channel response correction ratio, extracts the channel brightness and grayscale values ​​of the area to be calibrated, determines the correction rule index according to the correction ratio and the adjustment record, performs regional matching and adjustment, and generates a detection image calibration execution comparison table.

Images