Optical imaging data analysis method and system based on artificial intelligence

Abstract

The embodiment of the invention relates to the technical field of image data processing, in particular to an optical imaging data analysis method and system based on artificial intelligence, and provides an image analysis thought based on a liquid crystal imaging wavelet descriptor. The embodiment of the invention aims to improve the accuracy and efficiency of liquid crystal visual chart quality inspection. Meanwhile, through an innovative algorithm debugging strategy, the intelligent level of the automatic quality inspection system is further improved, and powerful technical support is provided for technical progress and industrial upgrading of the liquid crystal display industry.

Description

Technical Field

[0001] The present application relates to the field of image data processing technology, and more specifically, to an optical imaging data analysis method and system based on artificial intelligence. Background Art

[0002] With the rapid development of LCD display technology, the quality inspection of LCD eye charts has become a key link to ensure product performance and user experience. However, traditional quality inspection methods often rely on manual visual inspection, which is not only inefficient but also easily affected by human factors, making it difficult to ensure the accuracy and consistency of inspection. With the rise of artificial intelligence technology, the use of algorithms to automate the quality inspection of LCD eye charts has become an inevitable trend in the development of the industry.

[0003] However, existing automated quality inspection technologies often have problems such as low recognition accuracy and slow processing speed when facing complex and changeable LCD image features. The limitations of existing technologies are particularly obvious when identifying subtle defects and classifying similar features. In addition, due to the imaging characteristics of LCD eye charts, their image data contains a large amount of detailed information, which requires quality inspection algorithms to have not only efficient processing capabilities, but also accurate feature extraction and classification capabilities. Summary of the invention

[0004] In view of this, the present application provides an optical imaging data analysis method and system based on artificial intelligence.

[0005] The embodiment of the present application provides an optical imaging data analysis method based on artificial intelligence, which is applied to an optical imaging data analysis system. The method includes: Acquire a liquid crystal imaging image sample set corresponding to the optical image learning data set; the liquid crystal imaging image sample set includes a first liquid crystal imaging image sample and a second liquid crystal imaging image sample; the first liquid crystal imaging image sample carries a quality inspection compliance annotation, and the second liquid crystal imaging image sample carries a quality inspection defect annotation; Determine a liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging image sample in the liquid crystal imaging image sample set; Clustering is performed according to a plurality of the liquid crystal imaging wavelet descriptors corresponding to the liquid crystal imaging image sample set to obtain a plurality of wavelet description cluster sets; Determining a target wavelet description cluster set according to the liquid crystal imaging wavelet descriptors corresponding to each of the plurality of wavelet description cluster sets; Adjusting the annotations of the first liquid crystal imaging image sample corresponding to the target wavelet description cluster set into the quality inspection defect annotations to obtain a liquid crystal imaging image adjustment sample set; The initial image analysis algorithm is debugged according to the sample set adjusted according to the liquid crystal imaging diagram to obtain the target image analysis algorithm.

[0006] In some technical solutions, the clustering according to the plurality of liquid crystal imaging wavelet descriptors corresponding to the liquid crystal imaging pattern sample set to obtain a plurality of wavelet description cluster sets includes: determining the wavelet description commonality value of the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample according to the liquid crystal imaging wavelet descriptors corresponding to the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample respectively; clustering the plurality of wavelet description commonality values ​​corresponding to the liquid crystal imaging pattern sample set to obtain the plurality of wavelet description cluster sets; The step of determining the target wavelet description cluster set according to the liquid crystal imaging wavelet descriptors corresponding to the plurality of wavelet description cluster sets includes: determining the target wavelet description cluster set according to the wavelet description commonality values ​​corresponding to the plurality of wavelet description cluster sets.

[0007] In some technical solutions, determining the target wavelet description cluster set according to the liquid crystal imaging wavelet descriptors corresponding to the plurality of wavelet description cluster sets includes: determining the wavelet description cluster set including the imaging defect wavelet descriptor in the plurality of wavelet description cluster sets as the target wavelet description cluster set; Wherein, the method for determining the imaging defect wavelet descriptor includes: Acquire a past liquid crystal imaging atlas corresponding to the past optical image data set; the past liquid crystal imaging atlas includes a first past liquid crystal imaging image and a second past liquid crystal imaging image; Determine a liquid crystal imaging wavelet descriptor corresponding to each past liquid crystal imaging image in the past liquid crystal imaging atlas; Performing clustering according to a number of the past liquid crystal imaging wavelet descriptors corresponding to the past liquid crystal imaging atlas, to obtain a number of past wavelet description cluster sets; Determine the proportion of the second image corresponding to each past wavelet description cluster set; Determine the past wavelet description cluster sets whose proportion of the second imaging image is greater than the first set threshold as defect description cluster sets; The past liquid crystal imaging wavelet descriptor in the defect description cluster is determined as the imaging defect wavelet descriptor.

[0008] In some technical schemes, the clustering according to the several past liquid crystal imaging wavelet descriptors corresponding to the past liquid crystal imaging atlas to obtain several past wavelet description cluster sets includes: determining the past wavelet description commonality value of the first past liquid crystal imaging image and the second past liquid crystal imaging image according to the past liquid crystal imaging wavelet descriptors corresponding to each of the first past liquid crystal imaging image and the second past liquid crystal imaging image; clustering according to the several past wavelet description commonality values ​​corresponding to the past liquid crystal imaging atlas to obtain the several past wavelet description cluster sets; The method further includes: determining the commonality value of the sampled description in the defect description cluster set as a commonality measurement threshold; the commonality value of the sampled description in the defect description cluster set is less than the first wavelet description commonality value; the first wavelet description commonality value is the wavelet description commonality value in the defect description cluster set except the sampled description commonality value; The method of determining a target wavelet description cluster set according to the wavelet description commonality values ​​corresponding to each of the plurality of wavelet description cluster sets includes: determining a target wavelet description commonality value in each wavelet description cluster set; the target wavelet description commonality value in any wavelet description cluster set is less than a second wavelet description commonality value in any wavelet description cluster set; the second wavelet description commonality value in any wavelet description cluster set is a wavelet description commonality value in any wavelet description cluster set except the target wavelet description commonality value; and determining a wavelet description cluster set whose target wavelet description commonality value is not less than the commonality measurement threshold as the target wavelet description cluster set.

[0009] In some technical solutions, the method further includes: determining image semantic blocks corresponding to the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example respectively; determining pixel description commonality values ​​of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example according to the image semantic blocks corresponding to the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example respectively; The step of clustering a plurality of wavelet description common values ​​corresponding to the liquid crystal imaging pattern sample set to obtain the plurality of wavelet description cluster sets includes: clustering a plurality of wavelet description common values ​​and a plurality of pixel description common values ​​corresponding to the liquid crystal imaging pattern sample set to obtain the plurality of wavelet description cluster sets.

[0010] In some technical solutions, the method further includes: constructing a first mapping rule corresponding to the image semantic block in the liquid crystal imaging sample; the first mapping rule includes adjusting the pixel features of the object label into a first linear vector, adjusting the pixel features of the table label into a second linear vector, and adjusting the pixel features of the curve label into a third linear vector; The step of determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern sample in the liquid crystal imaging pattern sample set includes: mapping each liquid crystal imaging pattern sample in the liquid crystal imaging pattern sample set into a first liquid crystal imaging wavelet descriptor according to the first mapping rule; and determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern sample according to the first liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern sample.

[0011] In some technical schemes, the method further includes: constructing a second mapping rule corresponding to the image semantic block; the second mapping rule is to construct a semantic feature relationship network of the target scale; wherein the target scale corresponds to a set number of longitudinal distribution labels, the set number of longitudinal distribution labels includes a number of first longitudinal distribution labels and a number of second longitudinal distribution labels; each first longitudinal distribution label corresponds to an imaging object, and a linear vector in each first longitudinal distribution label represents whether the imaging object corresponding to each first longitudinal distribution label exists in the image semantic block; each second longitudinal distribution label corresponds to a non-imaging object pixel feature; the non-imaging object pixel feature includes a linear vector and a curve feature; the linear vector in each second longitudinal distribution label represents whether the non-imaging object pixel feature corresponding to each second longitudinal distribution label exists in the image semantic block; The method of determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern example according to the first liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern example comprises: mapping each liquid crystal imaging pattern example into a second liquid crystal imaging wavelet descriptor according to the second mapping rule; and determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern example according to the first liquid crystal imaging wavelet descriptor and the second liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern example.

[0012] In some technical solutions, the method further includes: constructing a third mapping rule corresponding to the image semantic block; the third mapping rule includes dividing the image semantic block into two local image semantic blocks according to the target pixel feature in the image semantic block; mapping the two local image semantic blocks into local liquid crystal imaging wavelet descriptors respectively according to the first mapping rule; and determining the average calculation results of the two local liquid crystal imaging wavelet descriptors to obtain the liquid crystal imaging wavelet descriptor; The method of determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern example according to the first liquid crystal imaging wavelet descriptor and the second liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern example comprises: mapping each liquid crystal imaging pattern example into a third liquid crystal imaging wavelet descriptor according to the third mapping rule; and determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern example according to the first liquid crystal imaging wavelet descriptor, the second liquid crystal imaging wavelet descriptor and the third liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern example.

[0013] In some technical solutions, the method further includes: constructing a fourth mapping rule corresponding to the image semantic block; the fourth mapping rule includes dividing the image semantic block into two local image semantic blocks according to the target pixel feature in the image semantic block; mapping the two local image semantic blocks into local liquid crystal imaging wavelet descriptors respectively according to the second mapping rule; determining the average calculation results of the two local liquid crystal imaging wavelet descriptors to obtain the liquid crystal imaging wavelet descriptor; The method of determining the liquid crystal imaging wavelet descriptors corresponding to the respective liquid crystal imaging pattern examples according to the first liquid crystal imaging wavelet descriptors, the second liquid crystal imaging wavelet descriptors and the third liquid crystal imaging wavelet descriptors corresponding to the respective liquid crystal imaging pattern examples comprises: mapping the respective liquid crystal imaging pattern examples into fourth liquid crystal imaging wavelet descriptors according to the fourth mapping rule; and determining the liquid crystal imaging wavelet descriptors corresponding to the respective liquid crystal imaging pattern examples according to the first liquid crystal imaging wavelet descriptors, the second liquid crystal imaging wavelet descriptors, the third liquid crystal imaging wavelet descriptors and the fourth liquid crystal imaging wavelet descriptors corresponding to the respective liquid crystal imaging pattern examples.

[0014] In some independent technical solutions, determining the pixel description commonality value of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example according to the image semantic blocks respectively corresponding to the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example includes: Determine a target linkage imaging atlas of the first liquid crystal imaging image sample and the second liquid crystal imaging image sample according to the image semantic blocks respectively corresponding to the first liquid crystal imaging image sample and the second liquid crystal imaging image sample; Determine the image semantic block dimensions corresponding to the first liquid crystal imaging image sample and the second liquid crystal imaging image sample respectively; Determine a first pixel description commonality value of the first liquid crystal imaging image sample and the second liquid crystal imaging image sample according to the image semantic block dimensions respectively corresponding to the first liquid crystal imaging image sample and the second liquid crystal imaging image sample and the target linkage imaging atlas; According to the first pixel description commonality value, a pixel description commonality value of the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample is determined.

[0015] In some independent technical solutions, determining the pixel description commonality value of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example according to the first pixel description commonality value includes: Determining linkage pixel features of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example according to image semantic blocks corresponding to the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example respectively; Determine a second pixel description commonality value of the first liquid crystal imaging image sample and the second liquid crystal imaging image sample according to the image semantic block dimensions corresponding to the first liquid crystal imaging image sample and the second liquid crystal imaging image sample and the linkage pixel feature; The pixel description commonality value of the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample is determined according to the first pixel description commonality value and the second pixel description commonality value.

[0016] In some independent technical solutions, determining the pixel description commonality value of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example according to the first pixel description commonality value and the second pixel description commonality value includes: Determine the proportion of linear vectors corresponding to the first liquid crystal imaging image sample and the second liquid crystal imaging image sample, respectively, according to the image semantic blocks corresponding to the first liquid crystal imaging image sample and the second liquid crystal imaging image sample, respectively; Determine a third pixel description commonality value of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example according to the linear vector proportions corresponding to the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example respectively; The pixel description commonality value of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example is determined according to the first pixel description commonality value, the second pixel description commonality value and the third pixel description commonality value.

[0017] In some technical solutions, the debugging of the initial image analysis algorithm according to the liquid crystal imaging image adjustment sample set to obtain the target image analysis algorithm includes: entering each liquid crystal imaging image adjustment sample in the liquid crystal imaging image adjustment sample set into the initial image analysis algorithm to obtain a semantic quantization vector corresponding to each liquid crystal imaging image adjustment sample; determining a quality inspection status prediction result corresponding to each liquid crystal imaging image adjustment sample based on the semantic quantization vector corresponding to each liquid crystal imaging image adjustment sample; adjusting an algorithm weight of the initial image analysis algorithm according to a difference between a quality inspection status prediction result corresponding to each liquid crystal imaging image adjustment sample and a quality inspection status authentication result corresponding to each liquid crystal imaging image adjustment sample to obtain the target image analysis algorithm; The method also includes: obtaining a liquid crystal imaging image to be analyzed of the optical image data to be analyzed; entering the liquid crystal imaging image to be analyzed into the target image analysis algorithm to obtain an imaging quality inspection point of view; if the imaging quality inspection point of view characterizes that the liquid crystal imaging image to be analyzed is a quality inspection defect point of view, determining the optical image data to be analyzed as substandard optical image data.

[0018] The present application also provides an optical imaging data analysis system, comprising: a memory for storing program instructions and data; a processor for coupling with the memory and executing instructions in the memory to implement the above method.

[0019] The present application also provides a computer storage medium, comprising instructions, which implement the above method when executed on a processor.

[0020] The embodiment of this application proposes an image analysis idea based on liquid crystal imaging wavelet descriptors. By constructing a professional liquid crystal imaging sample set and using wavelet descriptors for refined feature extraction, the embodiment of this application aims to improve the accuracy and efficiency of liquid crystal eye chart quality inspection. At the same time, through innovative algorithm debugging strategies, the embodiment of this application further improves the intelligence level of the automated quality inspection system, providing strong technical support for the technological progress and industrial upgrading of the liquid crystal display industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings herein are incorporated in and constitute a part of the specification, illustrate embodiments consistent with the present application, and together with the description, are used to explain the principles of the present application.

[0022] Figure 1 A schematic flowchart of the steps of an optical imaging data analysis method based on artificial intelligence provided in an embodiment of the present application.

[0023] Figure 2 A structural block diagram of an optical imaging data analysis system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0024] The technical solution in this application will be described below in conjunction with the accompanying drawings.

[0025] When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application. It should be noted that the terms "first", "second", etc. in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0026] The method embodiments provided in the embodiments of the present application can be executed in an optical imaging data analysis system, a computer device or a similar computing device. Taking the operation on the optical imaging data analysis system as an example, the optical imaging data analysis system may include one or more processors (the processor may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory for storing data. Optionally, the above optical imaging data analysis system may also include a transmission device for communication functions. It can be understood by those of ordinary skill in the art that the above structure is only illustrative and does not limit the structure of the above optical imaging data analysis system. For example, the optical imaging data analysis system may also include more or fewer components than those shown above, or have a configuration different from that shown above.

[0027] The memory can be used to store computer programs, for example, software programs and modules of application software, such as a computer program corresponding to an optical imaging data analysis method based on artificial intelligence in an embodiment of the present application. The processor executes various functional applications and data processing by running the computer program stored in the memory, that is, implementing the above method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the optical imaging data analysis system via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0028] The transmission device is used to receive or send data via a network. A specific example of the above-mentioned network may include a wireless network provided by a communication provider of the optical imaging data analysis system. In one example, the transmission device includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0029] In conjunction with the above, please refer to Figure 1 , Figure 1 It is a flow chart of an optical imaging data analysis method based on artificial intelligence provided in an embodiment of the present application. The method is applied to an optical imaging data analysis system and may further include steps 110 to 160.

[0030] In an exemplary application scenario of quality inspection of liquid crystal eye chart, the optical imaging data analysis system plays a vital role. The following are the detailed steps of the optical imaging data analysis system to implement the technical solution.

[0031] Step 110, obtaining a liquid crystal imaging image sample set corresponding to the optical image learning data set; the liquid crystal imaging image sample set includes a first liquid crystal imaging image sample and a second liquid crystal imaging image sample; the first liquid crystal imaging image sample carries a quality inspection compliance annotation, and the second liquid crystal imaging image sample carries a quality inspection defect annotation.

[0032] First, in step 110, the optical imaging data analysis system obtains an optical image learning data set from a database or an external storage device, and this data set is specifically for the imaging of a liquid crystal eye chart. The data set contains two liquid crystal imaging sample examples: the first liquid crystal imaging sample example is a display screen image that meets the quality inspection standard, and the second liquid crystal imaging sample example shows various quality inspection defects. These sample examples are accompanied by corresponding annotations, indicating whether they meet the standards or the specific defects that exist.

[0033] Based on the above, the optical image learning dataset is a data set specifically used to learn and train optical image processing and analysis models. In this application scenario of LCD eye chart quality inspection, this dataset contains a large number of optical images of LCD eye charts, which capture the visual performance of the display under various conditions. These data are carefully selected and annotated so that the machine learning algorithm can extract useful features from them, thereby improving the ability to judge the quality of LCD eye charts.

[0034] The LCD imaging sample set is a part of specific sample examples extracted from the optical image learning dataset, focusing on the imaging characteristics of LCD eye charts. This sample set includes different types of LCD eye chart images, including images of display screens that work normally and images of display screens with various quality problems, to help the algorithm learn and identify the quality status of LCD eye charts.

[0035] The first LCD imaging sample refers to the display screen images that are marked as qualified in the LCD imaging sample set. These images show the visual effects that the LCD eye chart should present under normal circumstances, without any obvious quality defects. They are an important reference for training algorithms to identify the normal state of the display screen. In contrast to the first LCD imaging sample, these sample examples show images of display screens with quality inspection defects. These defects may include bright spots, dark spots, color distortion, insufficient contrast and other problems. The existence of the second LCD imaging sample enables the algorithm to learn how to identify and classify these quality problems.

[0036] The quality inspection compliance annotations are the annotation information for the first LCD imaging sample, which clearly indicates that the display screen represented by these images has passed the quality inspection. These annotations are important guidance in the training process of machine learning algorithms, helping the algorithm understand which image features are signs of quality inspection compliance.

[0037] Corresponding to the QC compliance annotations, these annotations mark the specific QC defects present in the second LCD imaging sample. These annotations describe the type, location, and severity of the defects in detail, providing the algorithm with rich learning material, enabling it to accurately identify and classify various QC defects.

[0038] In detail, in step 110, the first task of the optical imaging data analysis system is to obtain an optical image learning data set specifically for the liquid crystal eye chart. This data set has been collected and organized before, and contains a large amount of liquid crystal eye chart image data. The system needs to extract a liquid crystal imaging sample set from the data set, and this sample set will serve as the basis for subsequent analysis and learning. The liquid crystal imaging sample set consists of two types of sample examples: a first liquid crystal imaging sample example and a second liquid crystal imaging sample example. The first liquid crystal imaging sample examples are those display screen images that are marked as meeting the quality inspection standards, which represent the visual effects of the liquid crystal eye chart under normal working conditions. These images can have a clear display effect without obvious quality defects. In contrast, the second liquid crystal imaging sample example shows display screen images with various quality inspection defects. These defects may include uneven display, color deviation, bad pixels, etc. The existence of these sample examples is crucial for training the system to identify and classify quality problems of the liquid crystal eye chart. In order to ensure the accuracy of subsequent analysis, each liquid crystal imaging sample example is accompanied by corresponding annotation information. For the first LCD imaging sample, these annotations clearly indicate that the display screen represented by the image has passed the quality inspection standard; while for the second LCD imaging sample, the annotations describe in detail the existing quality inspection defects and their specific circumstances. These annotation information will provide important reference for subsequent wavelet descriptor extraction, clustering, and algorithm debugging.

[0039] Step 120: Determine the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging image sample in the liquid crystal imaging image sample set.

[0040] Next, in step 120, the optical imaging data analysis system processes each of the liquid crystal imaging sample sets to extract their respective liquid crystal imaging wavelet descriptors. Wavelet descriptors are an important tool that can capture multi-scale features of images, which helps the system understand and analyze image content more accurately.

[0041] Based on the above content, the liquid crystal imaging wavelet descriptor is a feature vector extracted from the liquid crystal eye chart image through wavelet transform technology. This descriptor can capture the detailed changes of the image at different scales and directions, thereby effectively representing the key information such as texture, edge and shape of the liquid crystal eye chart image. Specifically, the liquid crystal imaging wavelet descriptor can be composed of a multi-dimensional feature vector, in which each dimension represents the response intensity of the image under a specific wavelet basis function.

[0042] For example, a typical liquid crystal imaging wavelet descriptor may contain multiple levels of wavelet coefficients, each of which is further subdivided into different directions and frequency bands. These coefficients reflect the energy distribution of the image in different frequencies and directions and are important quantitative indicators of image features. In practical applications, the liquid crystal imaging wavelet descriptor can be expressed as a high-dimensional numerical vector, such as [0.12, -0.05, 0.34, ..., -0.18], where each value represents the coefficient value after a specific wavelet transform.

[0043] Liquid crystal imaging wavelet descriptors play a key role in the quality inspection of liquid crystal eye charts, because they can help the algorithm accurately identify and classify various subtle defects on liquid crystal eye charts, such as bright spots, dark spots, line defects, etc. By comparing the wavelet descriptors of normal and defective displays, an efficient classification model can be trained to improve the automation and accuracy of quality inspection.

[0044] In detail, in step 120, the core task of the optical imaging data analysis system is to determine the wavelet descriptor corresponding to each sample example in the liquid crystal imaging sample set. This process is the basis for subsequent image analysis and defect detection, and is therefore crucial. First, the system traverses each sample example in the liquid crystal imaging sample set. For each sample example, the system applies a wavelet transform algorithm, which is a mathematical tool widely used in signal processing and can effectively extract multi-scale features of the image. When performing wavelet transform, the system selects appropriate wavelet basis functions, such as Haar wavelet, Daubechies wavelet, etc. These functions have good time-frequency localization characteristics and can capture detailed changes in the image. Through multi-level wavelet decomposition, the image is decomposed into sub-images of different scales and directions, each of which contains information of the original image in a specific frequency band and direction. Subsequently, the system calculates the wavelet coefficients of each sub-image, which reflect the energy distribution of the image in different frequencies and directions. In order to form a unified and representative feature vector, the system selects key wavelet coefficients and combines them into a high-dimensional numerical vector, namely, the liquid crystal imaging wavelet descriptor. This process is repeated for each sample in the LCD imaging sample set until all sample examples correspond to a unique wavelet descriptor. These descriptors not only capture the detailed features of the image, but also provide strong data support for subsequent image classification and defect detection. Through the processing of step 120, the optical imaging data analysis system successfully converts the original LCD eye chart image into a series of representative numerical feature vectors, laying a solid foundation for subsequent clustering, classification and algorithm debugging.

[0045] Step 130 , clustering is performed according to a plurality of the liquid crystal imaging wavelet descriptors corresponding to the liquid crystal imaging image sample set to obtain a plurality of wavelet description cluster sets.

[0046] Entering step 130, the system uses the extracted wavelet descriptors to perform cluster analysis on the image samples, that is, "clustering". Through a certain algorithm, the system can cluster images with similar features together to form several wavelet description cluster sets. The images within each cluster set have a high degree of similarity in features.

[0047] Based on the above content, the wavelet description cluster set is a data set obtained by extracting image features and performing cluster analysis using wavelet transform technology during the quality inspection of LCD eye charts. Wavelet transform is a method that can analyze local features of images at different scales and directions. Through wavelet transform, detailed information and structural features of images can be extracted. These features are called wavelet descriptors. In quality inspection applications, wavelet descriptors are used to characterize various attributes of LCD eye chart images, such as brightness, contrast, texture, etc. After the wavelet descriptors are extracted, the system uses clustering algorithms to divide them into different groups (or clusters) based on the similarities and differences of these descriptors. The images within each group have high similarity in features, while the images between different groups have great differences. The data set formed in this way is called a wavelet description cluster set. Each cluster set can be regarded as a collection of LCD eye chart images with similar features, which is crucial for subsequent defect detection and classification. The wavelet description cluster set can not only help quality inspectors identify and locate potential defects of LCD eye charts more quickly, but also provide strong data support for the debugging and optimization of image analysis algorithms.

[0048] In detail, step 130 is a key step in the optical imaging data analysis system, which involves clustering analysis of the liquid crystal imaging sample to form a wavelet-described cluster set. The following is a detailed description of this step.

[0049] (1) Data preparation: Before entering step 130, the system has completed the extraction of wavelet descriptors for the liquid crystal imaging sample. These descriptors capture the multi-scale features of the image and provide rich information for subsequent clustering analysis.

[0050] (2) Select a clustering algorithm: The system selects a suitable clustering algorithm for clustering based on actual needs and data characteristics. Commonly used clustering algorithms include K-means, hierarchical clustering, DBSCAN, etc. When selecting an algorithm, it is necessary to consider the efficiency, accuracy, and adaptability of the algorithm to data distribution.

[0051] (3) Perform clustering: Using the selected clustering algorithm, the system clusters the liquid crystal imaging wavelet descriptors. In this process, the algorithm groups the descriptors according to their similarity to form different clusters (i.e., sub-clusters). The descriptors within each cluster have high similarity, while the descriptors between clusters have great differences.

[0052] (4) Forming wavelet-described clusters: After clustering, the system classifies the liquid crystal imaging samples in each cluster into the corresponding wavelet-described clusters. In this way, each wavelet-described cluster represents a collection of images with some common features or attributes.

[0053] (5) Verification and optimization: In order to ensure the accuracy and effectiveness of clustering, the system can use some evaluation indicators (such as silhouette coefficient, Davies-Bouldin index, etc.) to evaluate the clustering results. If the clustering effect is not good, the system can adjust the parameters of the clustering algorithm or try other algorithms for optimization.

[0054] (6) Output and storage: Finally, the system outputs the formed wavelet description cluster set and stores it in a database or external storage device for use in subsequent steps.

[0055] Through the cluster analysis in step 130 , the optical imaging data analysis system can more effectively organize and utilize the liquid crystal imaging sample data, thereby providing strong support for subsequent quality inspection and defect identification.

[0056] Step 140 : determining a target wavelet description cluster set according to the liquid crystal imaging wavelet descriptors corresponding to each of the plurality of wavelet description cluster sets.

[0057] In step 140, the optical imaging data analysis system further analyzes these wavelet description clusters and determines a target wavelet description cluster according to the liquid crystal imaging wavelet descriptor therein. This target cluster may contain images of certain specific defects or features that the quality inspector is particularly concerned about.

[0058] Based on the above, the target wavelet description cluster set is a specific set selected from multiple wavelet description cluster sets after being processed and analyzed by the optical imaging data analysis system during the quality inspection of the LCD eye chart. The images in this set have similar wavelet descriptor features, which may point to defects or specific attributes that quality inspectors pay special attention to. The wavelet descriptor is a feature descriptor extracted from an image through wavelet transform, which can capture detailed information of the image at different scales and directions. The determination of the target wavelet description cluster set helps the quality inspection system to more accurately identify and classify various defects on the LCD eye chart, thereby improving the efficiency and accuracy of quality inspection.

[0059] In detail, step 140 is one of the key steps in the optical imaging data analysis system, which involves determining a target wavelet description cluster set from multiple wavelet description cluster sets. The following is a detailed introduction to this step.

[0060] (1) Analyzing wavelet description clusters: After completing step 130, the system has obtained several wavelet description clusters, and the images in each cluster have similar wavelet descriptor features. The first task of step 140 is to conduct in-depth analysis of these wavelet description clusters.

[0061] (2) Feature comparison and screening: The system compares the wavelet descriptors in each wavelet description cluster to find features that are highly correlated with quality inspection defects or specific attributes. This may involve comparing with pre-set standards or historical data to identify clusters that may contain important information.

[0062] (3) Determine the target cluster set: Based on the result of feature comparison, the system selects one or several wavelet description cluster sets that are most relevant to the quality inspection target as the target wavelet description cluster set. This set may contain images with certain defects or specific image attributes that the quality inspectors are particularly concerned about, and is the focus of subsequent quality inspection analysis.

[0063] (4) Verification and adjustment: After determining the target wavelet description cluster set, the system can perform a verification process to ensure that the selected cluster set does meet the quality inspection requirements. If it is found that the selected cluster set does not meet the expectations, the system will make corresponding adjustments and reselect or optimize the target wavelet description cluster set.

[0064] (5) Output and record: Finally, the system outputs the determined target wavelet description cluster set and records the relevant information for use in subsequent steps. This target cluster set will become an important reference in the subsequent quality inspection process, helping the quality inspection system to more accurately identify and classify defects on the LCD eye chart.

[0065] By accurately screening and determining the target wavelet description cluster set in step 140, the optical imaging data analysis system can conduct subsequent quality inspection analysis more specifically and improve the efficiency and accuracy of quality inspection. This is a key step for LCD eye chart manufacturers to ensure product quality and improve production efficiency.

[0066] Step 150: Adjust the annotations of the first liquid crystal imaging image sample corresponding to the target wavelet description cluster set into the quality inspection defect annotations to obtain a liquid crystal imaging image adjustment sample set.

[0067] After determining the target wavelet description cluster set, step 150 requires the system to adjust the annotations of the first LCD imaging image sample (the image originally marked as up to standard in quality inspection) in this specific cluster set. This is because in the actual quality inspection process, it can be found that some images that were originally considered up to standard actually have subtle defects. By adjusting the annotations of these images to quality inspection defect annotations, the system can create a more accurate LCD imaging image adjustment sample set.

[0068] Based on the above content, the LCD imaging image adjustment sample set refers to a set of images formed after specific processing and analysis during the quality inspection process of the LCD eye chart. The images in this set originally belonged to the first LCD imaging image sample, that is, the images that were originally marked as meeting the quality inspection standards. However, in further detailed analysis and comparison, it can be found that there are subtle defects in these images that were not detected during the initial quality inspection. Therefore, the labeling of these images needs to be adjusted from the original "quality inspection standards" to "quality inspection defects." Such an adjustment sample set is of great significance for optimizing and improving the quality inspection accuracy of LCD eye charts. It can help the quality inspection system better learn and identify various potential and difficult-to-detect display screen defects, thereby improving the level of product quality control.

[0069] In detail, in the quality inspection process of the liquid crystal eye chart, step 150 is a key link, which involves the adjustment of the annotation of the images in the target wavelet description cluster set, a specific image set. The purpose of this step is to more accurately reflect the actual situation in the image and further optimize the accuracy of the quality inspection system. Specifically, after the optical imaging data analysis system determines the target wavelet description cluster set through the previous steps, it will conduct a detailed review and analysis of the first liquid crystal imaging sample in this set. These first liquid crystal imaging samples were originally considered to be up to standard for quality inspection, but the system can find subtle defects hidden in them through in-depth data analysis and comparison. These defects may not be detected in the preliminary quality inspection, but for manufacturers pursuing high-quality products, they are issues that cannot be ignored. Therefore, in step 150, the system adjusts these newly discovered defective images from the original "quality inspection standard" annotation to the "quality inspection defect" annotation. This adjustment process is not only a correction of the preliminary quality inspection results, but also an improvement of the learning ability of the quality inspection system itself. Through this marked adjustment, the system can "remember" the characteristics of these subtle defects, so that similar problems can be more accurately identified in future quality inspections. Finally, after processing in step 150, an adjusted sample set of LCD imaging images will be obtained. The images in this set have been carefully reviewed and marked, and they reflect various subtle defects that may exist in the LCD eye chart. This adjusted sample set will become an important reference for subsequent quality inspections, helping manufacturers to continuously improve product quality and customer satisfaction.

[0070] Step 160 , adjusting the sample set according to the liquid crystal imaging image to debug the initial image analysis algorithm and obtain the target image analysis algorithm.

[0071] Finally, in step 160, the optical imaging data analysis system uses the adjusted sample set to debug and optimize the initial image analysis algorithm. Through continuous iteration and adjustment, the system can eventually obtain a more accurate and efficient target image analysis algorithm, which will be used to automatically detect the quality of LCD eye charts and accurately identify various potential defects in the future.

[0072] Based on the above, the initial image analysis algorithm refers to the image recognition and processing algorithm initially used in the quality inspection process of the LCD eye chart. This algorithm may be based on some common image processing technologies, such as edge detection, color analysis, shape recognition, etc., and is used to automatically or semi-automatically detect possible defects on the LCD eye chart. However, due to its versatility, this initial algorithm may not accurately identify certain specific types of defects, or may not work well when processing images with special textures and lighting conditions. Therefore, it needs to be debugged and optimized through actual application scenarios and data.

[0073] The target image analysis algorithm is an image analysis algorithm that has been debugged and optimized to better meet the quality inspection needs of specific LCD eye charts. Based on the initial image analysis algorithm, this algorithm makes targeted improvements to the initial algorithm by analyzing and learning a large amount of actual image data (such as the LCD imaging adjustment sample set). These improvements may include adjusting parameters, adding specific feature extraction methods, optimizing classifiers, etc. to improve the accuracy and efficiency of identifying specific defects. The target image analysis algorithm is the core algorithm of the optical imaging data analysis system that is ultimately used for automated quality inspection, and its performance directly affects the accuracy and efficiency of quality inspection.

[0074] In detail, in the final stage of the quality inspection of the LCD eye chart, the optical imaging data analysis system ushered in the critical step 160. The core task of this step is to debug and optimize the initial image analysis algorithm using the LCD imaging image adjustment sample set carefully constructed in the previous step to obtain a more accurate and efficient target image analysis algorithm. The system first loads the initial image analysis algorithm, which is a basic algorithm that already has certain image processing capabilities. However, this algorithm may not be fully adapted to the various complex situations and specific defects in the quality inspection of the LCD eye chart. Therefore, the system needs to fine-tune it through the LCD imaging image adjustment sample set. The LCD imaging image adjustment sample set contains various carefully annotated images that reflect various actual situations in the quality inspection process, including common and special defect types. The system inputs these images into the initial image analysis algorithm and observes the output results of the algorithm. By comparing with the annotated information in the sample set, the system can accurately evaluate the performance of the algorithm and find out its shortcomings. Next, the system gradually debugs and optimizes the initial image analysis algorithm based on the evaluation results. This may include adjusting the algorithm's parameter settings, adding new feature extraction methods to increase sensitivity to specific defects, or optimizing the performance of the classifier to reduce misjudgments and missed detections. This process may need to be repeated many times, and the sample set must be re-verified after each adjustment to ensure that the algorithm's performance is continuously improved. In the end, after a series of debugging and optimization work, the system obtained a target image analysis algorithm with significantly improved performance. This algorithm can not only more accurately identify various defects on LCD eye charts, but also achieve higher standards in processing speed and stability. In the future, this target image analysis algorithm will become a core component of the optical imaging data analysis system, providing strong support for the quality inspection of LCD eye charts.

[0075] It can be seen that in the whole process, the optical imaging data analysis system not only improves the efficiency and accuracy of quality inspection by automatically processing and analyzing a large amount of LCD eye chart image data, but also greatly reduces labor costs and misjudgment rate. This is undoubtedly a huge technological advancement for LCD eye chart manufacturers.

[0076] The embodiment of the present application not only realizes the refined detection of the quality of the liquid crystal eye chart by obtaining a specific set of liquid crystal imaging sample examples and using the liquid crystal imaging wavelet descriptor to extract and classify image features, but also further improves the intelligent level of quality inspection. The specific beneficial effects are shown in the following aspects: First, by constructing a set of liquid crystal imaging sample examples containing quality inspection compliance and quality inspection defect annotations, the embodiment of the present application provides a rich and accurate data basis for subsequent image analysis, ensuring the reliability of the analysis results. Secondly, the embodiment of the present application uses liquid crystal imaging wavelet descriptors to characterize image features. This method can deeply capture the detailed information of the image and provide strong support for subsequent image classification and recognition. Furthermore, through the cluster processing of wavelet descriptors, the embodiment of the present application can quickly locate a set of images with similar features, which not only improves the efficiency of image processing, but also provides the possibility for accurate identification of specific defects. In addition, the embodiment of the present application also innovatively proposes a strategy for adjusting the compliance image annotations in a specific cluster set to defect annotations. This step effectively corrects the errors that may exist in the initial annotations, further enhancing the accuracy and practicality of the data set. Finally, by debugging the initial image analysis algorithm using the adjusted sample set, the embodiment of the present application successfully obtained a more accurate and efficient target image analysis algorithm. In practical applications, the algorithm can quickly identify and classify objects, features, and anomalies in images, providing users with higher-quality decision support and significantly improving the automation and intelligence level of LCD eye chart quality inspection.

[0077] In some optional embodiments, the step of performing clustering according to the plurality of liquid crystal imaging wavelet descriptors corresponding to the liquid crystal imaging pattern sample set to obtain a plurality of wavelet description cluster sets includes: determining the wavelet description commonality value of the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample according to the liquid crystal imaging wavelet descriptors corresponding to the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample respectively; performing clustering on the plurality of wavelet description commonality values ​​corresponding to the liquid crystal imaging pattern sample set to obtain the plurality of wavelet description cluster sets. Then, the step of determining the target wavelet description cluster set according to the liquid crystal imaging wavelet descriptors corresponding to the plurality of wavelet description cluster sets includes: determining the target wavelet description cluster set according to the wavelet description commonality values ​​corresponding to the plurality of wavelet description cluster sets respectively.

[0078] In this embodiment, when the optical imaging data analysis system executes the technical solution, a series of refined operations will be performed. When the system performs clustering based on the liquid crystal imaging wavelet descriptors corresponding to the liquid crystal imaging sample set, the specific steps are as follows: First, the optical imaging data analysis system extracts the liquid crystal imaging wavelet descriptors corresponding to the first liquid crystal imaging sample and the second liquid crystal imaging sample, respectively. As a mathematical tool, the wavelet descriptor can accurately capture the characteristic changes of the image at different scales and directions, providing rich information for subsequent image analysis.

[0079] Next, the system calculates the wavelet description commonality value between the first liquid crystal imaging sample and the second liquid crystal imaging sample based on these wavelet descriptors. This commonality value is actually a quantitative index that reflects the similarity of different images in terms of features. The higher the commonality value, the closer the images are in terms of features.

[0080] Subsequently, the optical imaging data analysis system performs cluster analysis on all the calculated wavelet description commonality values, which is called "clustering". In this process, the system clusters images with similar commonality values ​​together to form several wavelet description cluster sets. Each cluster set represents a class of images with similar characteristics.

[0081] After determining these wavelet description clusters, the system further analyzes their corresponding wavelet description commonality values ​​to find the target wavelet description cluster that best meets the quality inspection requirements. This target cluster may contain images of certain features or defects that quality inspectors are particularly concerned about, which has important guiding significance for subsequent quality inspection work.

[0082] Through the above steps, the optical imaging data analysis system can realize the refined classification and processing of the liquid crystal imaging sample set, providing strong support for subsequent image analysis and quality inspection work.

[0083] In this way, by extracting the wavelet descriptors of liquid crystal imaging and performing refined cluster processing based on these descriptors, not only the efficiency of image processing is improved, but also the accuracy of classification is ensured. At the same time, by determining the target wavelet description cluster set, the system can quickly locate the image of specific features or defects that the quality inspectors are concerned about, providing strong data support for subsequent quality inspection decisions. The implementation of this technical solution has greatly improved the intelligent level of LCD eye chart quality inspection, reduced labor costs and misjudgment rate, and brought significant technological progress and industrial upgrading to the LCD display industry.

[0084] In some other optional technical solutions, the target wavelet description cluster set is determined according to the liquid crystal imaging wavelet descriptors corresponding to each of the plurality of wavelet description cluster sets, including: determining the wavelet description cluster set including the imaging defect wavelet descriptor in the plurality of wavelet description cluster sets as the target wavelet description cluster set; wherein the method for determining the imaging defect wavelet descriptor includes: acquiring a past liquid crystal imaging atlas corresponding to a past optical image data set; the past liquid crystal imaging atlas includes a first past liquid crystal imaging image and a second past liquid crystal imaging image; determining Determine the liquid crystal imaging wavelet descriptors corresponding to each past liquid crystal imaging image in the past liquid crystal imaging atlas; perform clustering according to the several past liquid crystal imaging wavelet descriptors corresponding to the past liquid crystal imaging atlas to obtain several past wavelet description cluster sets; determine the proportion of the second imaging image corresponding to each past wavelet description cluster set; determine the past wavelet description cluster set whose proportion of the second imaging image is greater than a first set threshold as a defect description cluster set; determine the past liquid crystal imaging wavelet descriptors in the defect description cluster set as the imaging defect wavelet descriptor.

[0085] In this technical solution, the optical imaging data analysis system adopts a specific method when determining the target wavelet description cluster set. This method first focuses on the wavelet description cluster sets that contain imaging defect wavelet descriptors, and these wavelet description cluster sets that are of special concern are determined as target wavelet description cluster sets.

[0086] In order to clarify what is the imaging defect wavelet descriptor, its determination process needs to be elaborated in detail. First, the system obtains a past liquid crystal imaging atlas corresponding to the past optical image data set, and this atlas contains at least two types of past liquid crystal imaging images: a first past liquid crystal imaging image and a second past liquid crystal imaging image. These two imaging images may represent different imaging conditions or qualities.

[0087] Next, the system determines the corresponding liquid crystal imaging wavelet descriptor for each image in the past liquid crystal imaging atlas. Wavelet descriptors are mathematical tools that can capture details and features in images and are essential for image recognition and analysis.

[0088] Then, according to these past liquid crystal imaging wavelet descriptors, the system performs clustering processing on them to form several past wavelet description cluster sets. Each cluster set represents a group of images with similar features.

[0089] In order to identify which clusters may contain imaging defects, the system further analyzes the proportion of the second past liquid crystal imaging image in each past wavelet description cluster. If the proportion of the second imaging image in a cluster exceeds the preset first set threshold, then the cluster is identified as a defect description cluster. The first set threshold in the embodiment of the present application is a key parameter, which is used to define when a cluster is considered to contain enough imaging defects to be paid special attention to.

[0090] Finally, the system identifies the past liquid crystal imaging wavelet descriptors in the defect description cluster as imaging defect wavelet descriptors. These descriptors represent the image features with significant imaging defects and are crucial for subsequent analysis and processing.

[0091] In this way, the optical imaging data analysis system can accurately identify and focus on image sets that contain imaging defects, providing strong data support for subsequent optimization and correction work. The implementation of this method not only improves the accuracy of image analysis, but also provides a scientific basis for improving the quality of LCD displays.

[0092] In a replaceable embodiment, clustering is performed according to several past liquid crystal imaging wavelet descriptors corresponding to the past liquid crystal imaging atlas to obtain several past wavelet description cluster sets, including: determining the past wavelet description commonality value of the first past liquid crystal imaging image and the second past liquid crystal imaging image according to the past liquid crystal imaging wavelet descriptors corresponding to each of the first past liquid crystal imaging image and the second past liquid crystal imaging image; clustering is performed according to several past wavelet description commonality values ​​corresponding to the past liquid crystal imaging atlas to obtain the several past wavelet description cluster sets.

[0093] The method also includes: determining the sampling description commonality value in the defect description cluster set as the commonality measurement threshold; the sampling description commonality value in the defect description cluster set is less than the first wavelet description commonality value; the first wavelet description commonality value is the wavelet description commonality value in the defect description cluster set except the sampling description commonality value.

[0094] Furthermore, determining the target wavelet description cluster set according to the wavelet description commonality values ​​corresponding to each of the plurality of wavelet description cluster sets includes: determining the target wavelet description commonality value of each wavelet description cluster set; the target wavelet description commonality value of any wavelet description cluster set is less than the second wavelet description commonality value of any wavelet description cluster set; the second wavelet description commonality value of any wavelet description cluster set is the wavelet description commonality value of any wavelet description cluster set except the target wavelet description commonality value; and determining the wavelet description cluster set whose target wavelet description commonality value is not less than the commonality measurement threshold as the target wavelet description cluster set.

[0095] In the application scenario of liquid crystal eye chart, the optical imaging data analysis system plays a vital role. In an alternative embodiment, the system performs in-depth analysis on past liquid crystal imaging atlases to identify and classify features in the image.

[0096] First, the system determines the wavelet description commonality value between the first past liquid crystal imaging image and the second past liquid crystal imaging image based on the wavelet descriptors corresponding to the two types of imaging images. This commonality value is actually a quantitative indicator used to measure the similarity or consistency of the two types of images in terms of wavelet features. By calculating these commonality values, the system can gain insight into the internal connections and differences between the images.

[0097] Next, the system performs clustering based on the calculated commonality values ​​of the past wavelet descriptions. "Clustering" in the embodiment of the present application refers to grouping image sets with similar commonality values ​​together to form several past wavelet description cluster sets. Each cluster set represents a group of images with a certain commonality of specific wavelet features.

[0098] In this process, the system also pays special attention to clusters that may contain imaging defects. In order to accurately identify these defects, the system determines the commonality value of the sampled descriptions in the defect description cluster as the commonality measurement threshold. This threshold value serves as a benchmark for the subsequent screening and classification of wavelet description clusters. It is worth noting that the commonality value of the sampled descriptions in the defect description cluster is less than the commonality value of other wavelet descriptions in the cluster (i.e., the commonality value of the first wavelet description), which reflects an abnormality or deviation in the imaging characteristics of the cluster.

[0099] Furthermore, when the system needs to determine the target wavelet description cluster set, it will first determine the target wavelet description commonality value in each wavelet description cluster set. This value is a specific value among all the wavelet description commonality values ​​in the cluster set, which is smaller than the other wavelet description commonality values ​​in the cluster set (i.e., the second wavelet description commonality value). Subsequently, the system determines those wavelet description cluster sets whose target wavelet description commonality values ​​are not less than the previously determined commonality measurement threshold as the target wavelet description cluster sets.

[0100] The implementation of this technical solution enables the optical imaging data analysis system to accurately identify and classify groups of images with specific wavelet feature commonalities, especially those that may contain imaging defects. In this way, the system not only improves the accuracy of image analysis, but also provides strong data support for subsequent image processing and optimization. In detail, through refined image feature analysis and classification, the performance of the optical imaging data analysis system in the application scenario of LCD eye charts has been significantly improved. It can accurately identify and focus on image sets with specific imaging characteristics, laying a solid foundation for improving LCD display quality and enhancing user experience.

[0101] In some other preferred embodiments, the method further includes: determining image semantic blocks corresponding to the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample respectively; and determining pixel description commonality values ​​of the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample based on the image semantic blocks corresponding to the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample respectively.

[0102] Then, clustering the plurality of wavelet description common values ​​corresponding to the liquid crystal imaging image sample set to obtain the plurality of wavelet description cluster sets includes: clustering the plurality of wavelet description common values ​​and the plurality of pixel description common values ​​corresponding to the liquid crystal imaging image sample set to obtain the plurality of wavelet description cluster sets.

[0103] In the application scenario of the liquid crystal eye chart, the optical imaging data analysis system further demonstrates its advanced functions. In other preferred embodiments, the system not only focuses on the wavelet descriptor, but also deeply mines the information of the image semantic block, so as to more comprehensively analyze the liquid crystal imaging sample.

[0104] First, the system determines the image semantic blocks corresponding to the first liquid crystal imaging sample and the second liquid crystal imaging sample. Image semantic blocks refer to areas in the image with specific semantics or meanings, such as a specific object, background or texture. By identifying these semantic blocks, the system can have a deeper understanding of the content and structure of the image.

[0105] Next, the system determines the pixel description commonality value of the first liquid crystal imaging sample and the second liquid crystal imaging sample based on these image semantic blocks. The pixel description commonality value is a quantitative indicator used to measure the similarity or consistency of two images at the pixel level. By calculating this value, the system can more accurately compare and analyze the subtle differences between different images.

[0106] Importantly, when performing clustering, the system not only considers the wavelet description commonality value, but also the pixel description commonality value. This means that the system performs joint clustering on several wavelet description commonality values ​​and several pixel description commonality values ​​corresponding to the liquid crystal imaging sample set, thereby obtaining a more accurate and detailed wavelet description cluster set.

[0107] This clustering method that combines wavelet descriptors and pixel description commonality values ​​enables the system to comprehensively analyze LCD imaging samples from multiple dimensions, improving the accuracy and effectiveness of clustering. In this way, the system can more accurately identify and classify image groups with similar features, providing more powerful support for subsequent image processing, analysis and optimization.

[0108] In this way, by integrating wavelet descriptors and pixel description commonality values ​​for cluster analysis, the optical imaging data analysis system has demonstrated higher efficiency and accuracy in the application scenario of LCD eye charts. This technical solution not only improves the depth of image analysis, but also provides a more solid technical foundation for improving LCD display quality and user experience.

[0109] In some preferred embodiments, the method also includes: constructing a first mapping rule corresponding to the image semantic block in the liquid crystal imaging sample; the first mapping rule includes adjusting the pixel features of the object label into a first linear vector, adjusting the pixel features of the table label into a second linear vector, and adjusting the pixel features of the curve label into a third linear vector.

[0110] The step of determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample in the liquid crystal imaging sample set includes: mapping each liquid crystal imaging sample in the liquid crystal imaging sample set into a first liquid crystal imaging wavelet descriptor according to the first mapping rule; and determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample according to the first liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample.

[0111] In some preferred embodiments, the optical imaging data analysis system further introduces a mapping rule to improve the analysis accuracy and efficiency of the liquid crystal imaging sample. In this process, the system first constructs a first mapping rule corresponding to the image semantic block in the liquid crystal imaging sample.

[0112] This first mapping rule specifically defines how to convert different types of image semantic blocks into corresponding linear vectors. According to the rule, the pixel features of the object label are adjusted into the first linear vector, the pixel features of the table label are adjusted into the second linear vector, and the pixel features of the curve label are adjusted into the third linear vector. The "object", "table" and "curve" labels here may be automatically identified based on the image content, or pre-defined to describe different types of elements or features in the image.

[0113] After determining the mapping rule, the system can map each liquid crystal imaging sample in the liquid crystal imaging sample set into the first liquid crystal imaging wavelet descriptor according to the mapping rule. This step actually converts the semantic information of the image into a form that is more convenient for mathematical processing and comparison, namely, the wavelet descriptor.

[0114] Specifically, the system traverses each sample in the liquid crystal imaging sample set, identifies the image semantic blocks therein, and converts these semantic blocks into corresponding linear vectors according to the first mapping rule. Subsequently, these linear vectors are further processed to generate the first liquid crystal imaging wavelet descriptor corresponding to each sample.

[0115] Finally, the system determines the liquid crystal imaging wavelet descriptors corresponding to each liquid crystal imaging sample according to these first liquid crystal imaging wavelet descriptors. This step may be to further optimize or standardize the first liquid crystal imaging wavelet descriptors to ensure that they can accurately reflect the characteristics of the image and facilitate subsequent analysis and comparison.

[0116] By introducing this mapping rule, the optical imaging data analysis system can more accurately capture and describe the features in the liquid crystal imaging sample, thereby improving the accuracy and efficiency of image analysis. The implementation of this technical solution not only helps the system to understand the image content more deeply, but also provides strong support for subsequent image processing, recognition and optimization. In other words, by constructing and applying the first mapping rule, the optical imaging data analysis system realizes the accurate analysis and description of the liquid crystal imaging sample in the application scenario of the liquid crystal eye chart. This technical solution significantly improves the intelligence and automation level of image analysis, and lays a solid foundation for improving the quality of liquid crystal display and enhancing user experience.

[0117] In other possible technical solutions, the method further includes: constructing a second mapping rule corresponding to the image semantic block; the second mapping rule is to construct a semantic feature relationship network of the target scale; wherein the target scale corresponds to a set number of longitudinal distribution labels, the set number of longitudinal distribution labels includes a number of first longitudinal distribution labels and a number of second longitudinal distribution labels; each first longitudinal distribution label corresponds to an imaging object, and the linear vector in each first longitudinal distribution label represents whether the imaging object corresponding to each first longitudinal distribution label exists in the image semantic block; each second longitudinal distribution label corresponds to a non-imaging object pixel feature; the non-imaging object pixel feature includes Linear vector and curve features; the linear vector in each second longitudinal distribution label represents whether there is a non-imaging object pixel feature corresponding to each second longitudinal distribution label in the image semantic block; the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample is determined based on the first liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample, including: mapping each liquid crystal imaging sample into a second liquid crystal imaging wavelet descriptor according to the second mapping rule; determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample based on the first liquid crystal imaging wavelet descriptor and the second liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample.

[0118] In this technical solution, the optical imaging data analysis system further introduces a second mapping rule to deepen the analysis of the liquid crystal imaging sample. This rule not only helps the system understand the image content more comprehensively, but also improves the recognition accuracy of image features.

[0119] Specifically, the second mapping rule is to construct a semantic feature relationship network of a target scale. The target scale of this relationship network corresponds to a set number of vertical distribution labels, which are divided into two categories: first vertical distribution labels and second vertical distribution labels.

[0120] The first longitudinal distribution label corresponds to the imaging object. Each first longitudinal distribution label represents a specific imaging object, such as a character, a graphic, etc. Among these labels, a linear vector is used to characterize whether there is an imaging object corresponding to the label in the image semantic block. In other words, the system checks the pixel features in the image semantic block to determine whether it matches a first longitudinal distribution label and represents the result in the form of a linear vector.

[0121] The second longitudinal distribution label corresponds to the pixel features of non-imaging objects. These features include linear vectors and curve features, which are used to describe the parts of the image that do not belong to a specific imaging object. Similar to the first longitudinal distribution label, the linear vector in the second longitudinal distribution label is also used to characterize whether there is a non-imaging object pixel feature corresponding to the label in the image semantic block.

[0122] After constructing the second mapping rule, the system can map each liquid crystal imaging sample into the second liquid crystal imaging wavelet descriptor according to the rule. This process involves in-depth analysis and feature extraction of image semantic blocks to ensure that the generated second liquid crystal imaging wavelet descriptor can accurately reflect various features in the image.

[0123] Subsequently, the system combines the first liquid crystal imaging wavelet descriptor and the second liquid crystal imaging wavelet descriptor to determine the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample. This step may be to fuse or weighted average the two types of descriptors to obtain a more comprehensive and accurate image feature description.

[0124] By introducing the second mapping rule and constructing a semantic feature relationship network, the optical imaging data analysis system can analyze various features in the LCD imaging sample more carefully, including the pixel features of the imaging objects and non-imaging objects. This not only improves the accuracy and depth of image analysis, but also provides a richer information basis for subsequent image processing, recognition and optimization. In this way, the optical imaging data analysis system realizes in-depth analysis and accurate description of the LCD imaging sample in the application scenario of the LCD eye chart. This technical solution significantly enhances the system's image analysis capabilities and provides strong support for improving the quality of LCD displays and user experience.

[0125] In the next step, the method also includes: constructing a third mapping rule corresponding to the image semantic block; the third mapping rule includes dividing the image semantic block into two local image semantic blocks according to the target pixel features in the image semantic block; mapping the two local image semantic blocks into local liquid crystal imaging wavelet descriptors respectively according to the first mapping rule; and determining the average calculation results of the two local liquid crystal imaging wavelet descriptors to obtain a liquid crystal imaging wavelet descriptor; determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample according to the first liquid crystal imaging wavelet descriptor and the second liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample, including: mapping each liquid crystal imaging sample into a third liquid crystal imaging wavelet descriptor respectively according to the third mapping rule; determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample according to the first liquid crystal imaging wavelet descriptor, the second liquid crystal imaging wavelet descriptor and the third liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample.

[0126] Based on this embodiment, the optical imaging data analysis system further introduces a third mapping rule to optimize the processing and analysis of the liquid crystal imaging sample. This rule is designed to improve the accuracy and representativeness of the liquid crystal imaging wavelet descriptor by refining the processing of the image semantic blocks.

[0127] The implementation process of the third mapping rule is as follows: First, the system divides the image semantic block into two local image semantic blocks based on the target pixel features in the image semantic block. The purpose of this step is to divide the image more finely in order to better capture and describe the local features. The target pixel features may include color, brightness, texture, etc. These features help the system determine how to effectively segment the image semantic block.

[0128] Next, the system maps the two local image semantic blocks into local liquid crystal imaging wavelet descriptors according to the first mapping rule. This step utilizes the multi-scale characteristics of wavelet transform and can capture local detail information of the image. By mapping local image semantic blocks into wavelet descriptors, the system can more accurately represent the characteristics of these areas.

[0129] Then, the system calculates the mean of the two local liquid crystal imaging wavelet descriptors to obtain a new liquid crystal imaging wavelet descriptor. The averaging calculation helps to reduce the influence of noise and outliers, making the final liquid crystal imaging wavelet descriptor more stable and reliable.

[0130] After applying the third mapping rule, the system can map each liquid crystal imaging sample into a third liquid crystal imaging wavelet descriptor according to the rule. These third liquid crystal imaging wavelet descriptors capture more local detail information in the image, providing richer data for subsequent image analysis and comparison.

[0131] Finally, the system combines the first liquid crystal imaging wavelet descriptor, the second liquid crystal imaging wavelet descriptor and the third liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample to determine the final liquid crystal imaging wavelet descriptor for each liquid crystal imaging sample. This process may be a weighted average or other form of fusion of the three types of descriptors to ensure that the final liquid crystal imaging wavelet descriptor can fully and accurately reflect the overall characteristics and local details of the image.

[0132] By introducing the third mapping rule, the optical imaging data analysis system can process and analyze liquid crystal imaging samples more finely, thereby improving the accuracy and completeness of image feature description. The implementation of this technical solution not only helps the system understand the image content more deeply, but also provides a more accurate data basis for subsequent image processing, recognition and optimization.

[0133] In some other replaceable embodiments, the method also includes: constructing a fourth mapping rule corresponding to the image semantic block; the fourth mapping rule includes dividing the image semantic block into two local image semantic blocks according to the target pixel features in the image semantic block; mapping the two local image semantic blocks into local liquid crystal imaging wavelet descriptors according to the second mapping rule; determining the average calculation results of the two local liquid crystal imaging wavelet descriptors to obtain the liquid crystal imaging wavelet descriptor; determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample example according to the first liquid crystal imaging wavelet descriptor, the second liquid crystal imaging wavelet descriptor and the third liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample example, including: mapping each liquid crystal imaging sample example into a fourth liquid crystal imaging wavelet descriptor according to the fourth mapping rule; determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample example according to the first liquid crystal imaging wavelet descriptor, the second liquid crystal imaging wavelet descriptor, the third liquid crystal imaging wavelet descriptor and the fourth liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample example.

[0134] Based on this embodiment, the optical imaging data analysis system adopts another strategy to further enhance the analysis capability of the liquid crystal imaging sample, that is, by constructing and applying the fourth mapping rule. This rule is different from the previous mapping rule, and it combines some characteristics of the second mapping rule and processes the image semantic block more carefully.

[0135] Specifically, the fourth mapping rule first divides the image semantic block into two local image semantic blocks based on the target pixel features in the image semantic block. The purpose of this step is to conduct a deeper analysis of the local features of the image. The target pixel features may involve color distribution, brightness changes, or specific texture patterns, which provide a basis for the reasonable segmentation of the image.

[0136] Next, the system uses the second mapping rule to map the two segmented local image semantic blocks into local liquid crystal imaging wavelet descriptors. Similar to the second mapping rule, this step also involves building a semantic feature relationship network, but this time it is applied to a smaller image area, so that more detailed feature information can be captured.

[0137] Then, the system calculates the mean of the two local liquid crystal imaging wavelet descriptors to obtain a unified liquid crystal imaging wavelet descriptor. Through the averaging process, the system can reduce the noise and anomalies in the local features and obtain a more stable and representative descriptor.

[0138] After applying the fourth mapping rule, the optical imaging data analysis system can map each liquid crystal imaging sample into the fourth liquid crystal imaging wavelet descriptor. These descriptors not only contain the overall characteristics of the image, but also highlight the key local details, providing richer information for image analysis and comparison.

[0139] Finally, the system combines the first liquid crystal imaging wavelet descriptor, the second liquid crystal imaging wavelet descriptor, the third liquid crystal imaging wavelet descriptor and the newly generated fourth liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample to determine the final liquid crystal imaging wavelet descriptor for each liquid crystal imaging sample. This process may be achieved through weighted averaging, feature fusion or other advanced algorithms to ensure that the final descriptor can fully and accurately reflect all important features of the image.

[0140] By introducing the fourth mapping rule, the optical imaging data analysis system realizes multi-level and multi-angle analysis of liquid crystal imaging samples in the liquid crystal vision chart application scenario. The implementation of the embodiment of the present application not only improves the accuracy of image feature extraction, but also provides more detailed and accurate data support for subsequent image processing, recognition and optimization.

[0141] In summary, by combining the fourth mapping rule with previous technical means, the optical imaging data analysis system significantly enhances its image analysis capabilities in the application scenario of LCD eye charts. This comprehensive solution can more comprehensively capture and describe the characteristics of LCD imaging samples, laying a more solid foundation for improving LCD display quality and enhancing user experience.

[0142] In some independent embodiments, determining the pixel description commonality value of the first liquid crystal imaging example and the second liquid crystal imaging example based on the image semantic blocks respectively corresponding to the first liquid crystal imaging example and the second liquid crystal imaging example includes: determining the target linkage imaging atlas of the first liquid crystal imaging example and the second liquid crystal imaging example based on the image semantic blocks respectively corresponding to the first liquid crystal imaging example and the second liquid crystal imaging example; determining the image semantic block dimensions respectively corresponding to the first liquid crystal imaging example and the second liquid crystal imaging example; determining the first pixel description commonality value of the first liquid crystal imaging example and the second liquid crystal imaging example based on the image semantic block dimensions respectively corresponding to the first liquid crystal imaging example and the second liquid crystal imaging example and the target linkage imaging atlas; determining the pixel description commonality value of the first liquid crystal imaging example and the second liquid crystal imaging example based on the first pixel description commonality value.

[0143] Based on this embodiment, the optical imaging data analysis system adopts a unique method to determine the commonality value of pixel description between two liquid crystal imaging image samples (i.e., the first liquid crystal imaging image sample and the second liquid crystal imaging image sample). This method not only considers the overall characteristics of the image, but also deeply analyzes the similarity and correlation between the images.

[0144] First, the system determines the target linkage imaging atlas between the first liquid crystal imaging sample and the second liquid crystal imaging sample based on the image semantic blocks corresponding to the two samples. The target linkage imaging atlas can be understood as a collection of similar or related parts in the two sample examples, which reflects the commonality of the two sample examples at the pixel level.

[0145] Next, the system determines the dimensions of the image semantic blocks corresponding to the two LCD imaging samples. Dimension here refers to the number of features or parameters required to describe the image semantic block, which represents the complexity and amount of information in the image. By comparing the dimensions of the two image semantic blocks, the system can preliminarily assess the degree of similarity between them.

[0146] Then, based on the image semantic block dimensions of the two sample examples and the previously determined target linkage imaging atlas, the system calculates the first pixel description commonality value. This value is a quantitative indicator used to measure the similarity or correlation between the two sample examples at the pixel level. The calculation process may involve complex mathematical operations and image processing techniques to ensure the accuracy and reliability of the results.

[0147] Finally, based on the first pixel description commonality value, the system determines the pixel description commonality value of the first liquid crystal imaging sample and the second liquid crystal imaging sample. This final value is a comprehensive evaluation of the overall similarity of the two sample examples, which takes into account multiple aspects and features of the image and is therefore of high reference value.

[0148] Through this method, the optical imaging data analysis system can accurately evaluate the commonality values ​​of pixel descriptions between two liquid crystal imaging samples. This not only helps the system to understand the image content more accurately, but also provides strong support for subsequent image processing, analysis and comparison. It can be seen that this embodiment achieves a comprehensive and quantitative evaluation of the similarity between two liquid crystal imaging samples by determining the target linkage imaging atlas, comparing the image semantic block dimensions, and calculating the commonality values ​​of pixel descriptions. The embodiment of the present application significantly improves the accuracy and objectivity of image analysis, and provides a strong technical guarantee for image processing in the application scenario of liquid crystal eye charts.

[0149] In some other independent embodiments, determining the pixel description commonality value of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example based on the first pixel description commonality value includes: determining the linkage pixel characteristics of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example based on the image semantic blocks corresponding to each of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example; determining the second pixel description commonality value of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example based on the image semantic block dimensions corresponding to each of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example and the linkage pixel characteristics; determining the pixel description commonality value of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example based on the first pixel description commonality value and the second pixel description commonality value.

[0150] In this embodiment, the optical imaging data analysis system further refines the method for determining the commonality value of pixel descriptions between two liquid crystal imaging pattern examples (i.e., the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example). This method obtains a more comprehensive and accurate commonality value by deeply analyzing the pixel features and dimensions of the pattern examples.

[0151] First, the system identifies and extracts the linked pixel features between the first and second liquid crystal imaging patterns based on the image semantic blocks corresponding to the two patterns. Linked pixel features refer to pixel features that appear simultaneously in the two patterns and have similarities or associations, such as the same color, texture, or shape. The existence of these features indicates that the two patterns have certain commonalities at the pixel level.

[0152] Next, the system calculates the second pixel description commonality value based on the image semantic block dimensions of the two sample examples and the extracted linked pixel features. This value takes into account the image dimensional information and pixel features, and can more comprehensively reflect the similarity or correlation between the two sample examples. The calculation process may involve weighting the dimensions and features or other complex mathematical operations to ensure the accuracy of the results.

[0153] Finally, the system combines the first pixel description commonality value and the second pixel description commonality value calculated previously to comprehensively determine the pixel description commonality value between the first liquid crystal imaging sample and the second liquid crystal imaging sample. This comprehensive commonality value is a comprehensive evaluation of the similarity of the two sample examples at the pixel level. It takes into account multiple aspects and features of the image at the same time, so it has higher accuracy and reliability.

[0154] Through this method, the optical imaging data analysis system can more deeply analyze and compare the similarities and correlations between the two liquid crystal imaging samples. This not only helps the system to understand the image content more accurately, but also provides a more accurate data basis for subsequent image processing, recognition and optimization. It can be seen that this embodiment achieves an in-depth analysis and quantitative evaluation of the similarities between the two liquid crystal imaging samples by identifying the linked pixel features, calculating the second pixel description commonality value, and comprehensively determining the pixel description commonality value. The embodiment of the present application further improves the accuracy and comprehensiveness of image analysis, and provides more powerful technical support for image processing in the application scenario of liquid crystal vision charts.

[0155] In some other independent embodiments, determining the pixel description commonality value of the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample based on the first pixel description commonality value and the second pixel description commonality value includes: determining the proportion of linear vectors corresponding to the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample based on the image semantic blocks corresponding to each of the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample; determining the third pixel description commonality value of the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample based on the proportion of linear vectors corresponding to each of the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample; determining the pixel description commonality value of the first liquid crystal imaging pattern sample and the second liquid crystal imaging pattern sample based on the first pixel description commonality value, the second pixel description commonality value and the third pixel description commonality value.

[0156] Based on this embodiment, the optical imaging data analysis system adopts a comprehensive method to determine the commonality value of pixel description between two liquid crystal imaging pattern examples (i.e., the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example). This method combines multiple features of the image to obtain a more accurate and comprehensive commonality value.

[0157] First, the system determines the linear vector ratios corresponding to the first and second liquid crystal imaging samples based on the image semantic blocks corresponding to the two samples. The linear vector ratio can be understood as the weight or proportion of each feature vector in the image, which reflects the importance of different features in the image. By calculating the linear vector ratio, the system can more accurately grasp the key features of the image.

[0158] Next, the system determines the third pixel description commonality value based on the linear vector ratio of the two sample examples. This value takes into account the weight and proportion of the image feature vector and can reflect the similarity or difference between the two sample examples in key features. The calculation process may involve complex mathematical operations and feature comparisons to ensure the accuracy of the results.

[0159] Finally, the system combines the previously calculated first pixel description commonality value, the second pixel description commonality value, and the newly obtained third pixel description commonality value to determine the pixel description commonality value between the first liquid crystal imaging sample and the second liquid crystal imaging sample. This comprehensive commonality value is a comprehensive evaluation of the similarity of the two sample examples at multiple levels. It takes into account multiple factors such as the image dimension, features, and feature vector ratio, so it has higher accuracy and credibility.

[0160] Through this method, the optical imaging data analysis system can more accurately evaluate the similarity between two liquid crystal imaging samples. This not only helps the system to better understand the image content, but also provides more accurate data support for subsequent image processing, comparison and optimization. It can be seen that this embodiment achieves a comprehensive and accurate evaluation of the similarity between two liquid crystal imaging samples by calculating the proportion of linear vectors, determining the commonality value described by the third pixel, and integrating multiple commonality values. The embodiment of the present application further improves the accuracy and credibility of image analysis, and provides a more reliable technical guarantee for image processing in the application scenario of liquid crystal eye charts.

[0161] In some preferred embodiments, debugging the initial image analysis algorithm according to the liquid crystal imaging image adjustment sample set to obtain the target image analysis algorithm includes: entering each liquid crystal imaging image adjustment sample in the liquid crystal imaging image adjustment sample set into the initial image analysis algorithm to obtain a semantic quantization vector corresponding to each liquid crystal imaging image adjustment sample; determining a quality inspection status prediction result corresponding to each liquid crystal imaging image adjustment sample based on the semantic quantization vector corresponding to each liquid crystal imaging image adjustment sample; adjusting the algorithm weight of the initial image analysis algorithm according to the difference between the quality inspection status prediction result corresponding to each liquid crystal imaging image adjustment sample and the quality inspection status authentication result corresponding to each liquid crystal imaging image adjustment sample to obtain the target image analysis algorithm.

[0162] Furthermore, the method also includes: obtaining a liquid crystal imaging image to be analyzed of the optical image data to be analyzed; entering the liquid crystal imaging image to be analyzed into the target image analysis algorithm to obtain an imaging quality inspection point of view; if the imaging quality inspection point of view characterizes that the liquid crystal imaging image to be analyzed is a quality inspection defect point of view, determining the optical image data to be analyzed as substandard optical image data.

[0163] Based on this embodiment, the optical imaging data analysis system adopts a refined method to optimize and adjust the image analysis algorithm so as to more accurately identify and analyze the quality of the liquid crystal imaging image. In this process, the system first uses the liquid crystal imaging image adjustment sample set to debug the initial image analysis algorithm, and then obtains a more accurate target image analysis algorithm.

[0164] Specifically, the system enters each LCD image adjustment sample in the LCD image adjustment sample set into the initial image analysis algorithm. The purpose of this step is to allow the algorithm to initially understand and recognize the image features of these samples. Through the processing of the algorithm, the system obtains the semantic quantization vector corresponding to each LCD image adjustment sample. These vectors are numerical representations of image features, which help the system to more accurately analyze and compare the similarities and differences between different images.

[0165] Next, the system determines the quality inspection status prediction results corresponding to each LCD imaging adjustment sample based on these semantic quantization vectors. This prediction result is a preliminary assessment of the current image quality based on the algorithm. However, this prediction result may not be completely accurate, so it needs to be compared with the actual quality inspection status.

[0166] In order to improve the accuracy of the algorithm, the system compares the quality inspection status prediction results corresponding to each LCD imaging adjustment sample with the actual quality inspection status certification results. By comparing the difference between the two, the system can find out where the algorithm has deviations in judgment and adjust the algorithm weight of the initial image analysis algorithm accordingly. This adjustment process is to make the algorithm better adapt to the characteristics of the LCD imaging image, thereby improving the accuracy of quality inspection.

[0167] After the above steps, the system finally obtains an optimized target image analysis algorithm, which has higher accuracy and reliability in identifying and analyzing liquid crystal imaging images.

[0168] Furthermore, after obtaining the target image analysis algorithm, the system also applies it to the actual image processing task. Specifically, the system first obtains the liquid crystal imaging images to be analyzed in the optical image data to be analyzed. Then, these liquid crystal imaging images to be analyzed are entered into the target image analysis algorithm to obtain the corresponding imaging quality inspection viewpoints.

[0169] If the imaging quality inspection indicates that the analyzed LCD image is defective, that is, the algorithm determines that the image has quality problems, the system will determine the corresponding optical image data to be analyzed as substandard optical image data. This determination result provides an important basis for subsequent image screening, correction or re-collection, ensuring the controllability and consistency of image quality.

[0170] In summary, by using the LCD imaging image adjustment sample set to debug and optimize the initial image analysis algorithm, and applying the optimized algorithm to actual image processing tasks, the optical imaging data analysis system has achieved precise control of the quality of the LCD imaging image. This not only improves the accuracy of image analysis, but also provides a more reliable and efficient image processing solution for application scenarios such as LCD eye charts.

[0171] Further, Figure 2 The structural block diagram of the optical imaging data analysis system 300 is shown, including: a memory 310 for storing program instructions and data; a processor 320 for coupling with the memory 310 and executing the instructions in the memory 310 to implement the above method.

[0172] Furthermore, a computer storage medium is provided, comprising instructions, which implement the above method when executed on a processor.

[0173] In several embodiments provided in the embodiments of the present application, it should be understood that the disclosed device and method can also be implemented in other ways. The device and method embodiments described above are merely schematic, for example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the device, method and computer program product according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0174] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0175] If the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code. It should be noted that in this article, the term "include", "include" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also includes elements inherent to such process, method, article or device. Without more constraints, an element defined by the phrase "comprising a..." does not exclude the existence of other identical elements in the process, method, article or apparatus comprising the element.

[0176] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An optical imaging data analysis method based on artificial intelligence, characterized in that: Applied to an optical imaging data analysis system, the method comprises: Acquire a liquid crystal imaging image sample set corresponding to the optical image learning data set; the liquid crystal imaging image sample set includes a first liquid crystal imaging image sample and a second liquid crystal imaging image sample; the first liquid crystal imaging image sample carries a quality inspection compliance annotation, and the second liquid crystal imaging image sample carries a quality inspection defect annotation; Determine a liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging image sample in the liquid crystal imaging image sample set; Clustering is performed according to a plurality of the liquid crystal imaging wavelet descriptors corresponding to the liquid crystal imaging image sample set to obtain a plurality of wavelet description cluster sets; Determining a target wavelet description cluster set according to the liquid crystal imaging wavelet descriptors corresponding to each of the plurality of wavelet description cluster sets; Adjusting the annotations of the first liquid crystal imaging image sample corresponding to the target wavelet description cluster set into the quality inspection defect annotations to obtain a liquid crystal imaging image adjustment sample set; The initial image analysis algorithm is debugged according to the sample set adjusted according to the liquid crystal imaging diagram to obtain the target image analysis algorithm.

2. The method according to claim 1, characterized in that The step of performing clustering according to the plurality of liquid crystal imaging wavelet descriptors corresponding to the liquid crystal imaging image sample set to obtain a plurality of wavelet description cluster sets comprises: determining the wavelet description commonality value of the first liquid crystal imaging image sample and the second liquid crystal imaging image sample according to the liquid crystal imaging wavelet descriptors corresponding to the first liquid crystal imaging image sample and the second liquid crystal imaging image sample respectively; performing clustering on the plurality of wavelet description commonality values ​​corresponding to the liquid crystal imaging image sample set to obtain the plurality of wavelet description cluster sets; The step of determining the target wavelet description cluster set according to the liquid crystal imaging wavelet descriptors corresponding to the plurality of wavelet description cluster sets includes: determining the target wavelet description cluster set according to the wavelet description commonality values ​​corresponding to the plurality of wavelet description cluster sets.

3. The method according to claim 2, characterized in that The step of determining the target wavelet description cluster set according to the liquid crystal imaging wavelet descriptors corresponding to the plurality of wavelet description cluster sets includes: determining the wavelet description cluster set including the imaging defect wavelet descriptor in the plurality of wavelet description cluster sets as the target wavelet description cluster set; Wherein, the method for determining the imaging defect wavelet descriptor includes: Acquire a past liquid crystal imaging atlas corresponding to the past optical image data set; the past liquid crystal imaging atlas includes a first past liquid crystal imaging image and a second past liquid crystal imaging image; Determine a liquid crystal imaging wavelet descriptor corresponding to each past liquid crystal imaging image in the past liquid crystal imaging atlas; Performing clustering according to a number of the past liquid crystal imaging wavelet descriptors corresponding to the past liquid crystal imaging atlas, to obtain a number of past wavelet description cluster sets; Determine the proportion of the second image corresponding to each past wavelet description cluster set; Determine the past wavelet description cluster sets whose proportion of the second imaging image is greater than the first set threshold as defect description cluster sets; The past liquid crystal imaging wavelet descriptor in the defect description cluster is determined as the imaging defect wavelet descriptor.

4. The method according to claim 3, characterized in that The clustering according to the several past liquid crystal imaging wavelet descriptors corresponding to the past liquid crystal imaging atlas to obtain several past wavelet description cluster sets includes: determining the past wavelet description commonality value of the first past liquid crystal imaging image and the second past liquid crystal imaging image according to the past liquid crystal imaging wavelet descriptors respectively corresponding to the first past liquid crystal imaging image and the second past liquid crystal imaging image; clustering according to the several past wavelet description commonality values ​​corresponding to the past liquid crystal imaging atlas to obtain the several past wavelet description cluster sets; The method further includes: determining the commonality value of the sampled description in the defect description cluster set as a commonality measurement threshold; the commonality value of the sampled description in the defect description cluster set is less than the first wavelet description commonality value; the first wavelet description commonality value is the wavelet description commonality value in the defect description cluster set except the sampled description commonality value; The method of determining a target wavelet description cluster set according to the wavelet description commonality values ​​corresponding to each of the plurality of wavelet description cluster sets includes: determining a target wavelet description commonality value in each wavelet description cluster set; the target wavelet description commonality value in any wavelet description cluster set is less than a second wavelet description commonality value in any wavelet description cluster set; the second wavelet description commonality value in any wavelet description cluster set is a wavelet description commonality value in any wavelet description cluster set except the target wavelet description commonality value; and determining a wavelet description cluster set whose target wavelet description commonality value is not less than the commonality measurement threshold as the target wavelet description cluster set.

5. The method according to claim 2, characterized in that: The method further includes: determining image semantic blocks corresponding to the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example respectively; determining pixel description commonality values ​​of the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example according to the image semantic blocks corresponding to the first liquid crystal imaging pattern example and the second liquid crystal imaging pattern example respectively; The step of clustering a plurality of wavelet description common values ​​corresponding to the liquid crystal imaging pattern sample set to obtain the plurality of wavelet description cluster sets includes: clustering a plurality of wavelet description common values ​​and a plurality of pixel description common values ​​corresponding to the liquid crystal imaging pattern sample set to obtain the plurality of wavelet description cluster sets.

6. The method according to claim 2, characterized in that The method further includes: constructing a first mapping rule corresponding to an image semantic block in a liquid crystal imaging sample; the first mapping rule includes adjusting pixel features of an object label into a first linear vector, adjusting pixel features of a table label into a second linear vector, and adjusting pixel features of a curve label into a third linear vector; The step of determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern sample in the liquid crystal imaging pattern sample set includes: mapping each liquid crystal imaging pattern sample in the liquid crystal imaging pattern sample set into a first liquid crystal imaging wavelet descriptor according to the first mapping rule; and determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern sample according to the first liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern sample.

7. The method according to claim 6, characterized in that The method also includes: constructing a second mapping rule corresponding to the image semantic block; the second mapping rule is to construct a semantic feature relationship network of the target scale; wherein the target scale corresponds to a set number of longitudinal distribution labels, the set number of longitudinal distribution labels include a number of first longitudinal distribution labels and a number of second longitudinal distribution labels; each first longitudinal distribution label corresponds to an imaging object, and a linear vector in each first longitudinal distribution label represents whether the imaging object corresponding to each first longitudinal distribution label exists in the image semantic block; each second longitudinal distribution label corresponds to a non-imaging object pixel feature; the non-imaging object pixel feature includes a linear vector and a curve feature; the linear vector in each second longitudinal distribution label represents whether the non-imaging object pixel feature corresponding to each second longitudinal distribution label exists in the image semantic block; The method of determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern example according to the first liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern example comprises: mapping each liquid crystal imaging pattern example into a second liquid crystal imaging wavelet descriptor according to the second mapping rule; and determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern example according to the first liquid crystal imaging wavelet descriptor and the second liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging pattern example; Wherein, the method further includes: constructing a third mapping rule corresponding to the image semantic block; the third mapping rule includes dividing the image semantic block into two local image semantic blocks according to the target pixel features in the image semantic block; mapping the two local image semantic blocks into local liquid crystal imaging wavelet descriptors respectively according to the first mapping rule; and determining the average calculation results of the two local liquid crystal imaging wavelet descriptors to obtain the liquid crystal imaging wavelet descriptor; the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample example is determined according to the first liquid crystal imaging wavelet descriptor and the second liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample example, including: mapping each liquid crystal imaging sample example into a third liquid crystal imaging wavelet descriptor respectively according to the third mapping rule; determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample example according to the first liquid crystal imaging wavelet descriptor, the second liquid crystal imaging wavelet descriptor and the third liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample example; The method further includes: constructing a fourth mapping rule corresponding to the image semantic block; the fourth mapping rule includes dividing the image semantic block into two local image semantic blocks according to the target pixel features in the image semantic block; mapping the two local image semantic blocks into local liquid crystal imaging wavelet descriptors according to the second mapping rule; determining the average calculation results of the two local liquid crystal imaging wavelet descriptors to obtain a liquid crystal imaging wavelet descriptor; determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample according to the first liquid crystal imaging wavelet descriptor, the second liquid crystal imaging wavelet descriptor and the third liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample, including: mapping each liquid crystal imaging sample into a fourth liquid crystal imaging wavelet descriptor according to the fourth mapping rule; determining the liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample according to the first liquid crystal imaging wavelet descriptor, the second liquid crystal imaging wavelet descriptor, the third liquid crystal imaging wavelet descriptor and the fourth liquid crystal imaging wavelet descriptor corresponding to each liquid crystal imaging sample.

8. The method according to claim 1, characterized in that The debugging of the initial image analysis algorithm according to the liquid crystal imaging image adjustment sample set to obtain the target image analysis algorithm includes: entering each liquid crystal imaging image adjustment sample in the liquid crystal imaging image adjustment sample set into the initial image analysis algorithm to obtain a semantic quantization vector corresponding to each liquid crystal imaging image adjustment sample; determining a quality inspection status prediction result corresponding to each liquid crystal imaging image adjustment sample according to the semantic quantization vector corresponding to each liquid crystal imaging image adjustment sample; adjusting an algorithm weight of the initial image analysis algorithm according to a difference between a quality inspection status prediction result corresponding to each liquid crystal imaging image adjustment sample and a quality inspection status authentication result corresponding to each liquid crystal imaging image adjustment sample to obtain the target image analysis algorithm; The method also includes: obtaining a liquid crystal imaging image to be analyzed of the optical image data to be analyzed; entering the liquid crystal imaging image to be analyzed into the target image analysis algorithm to obtain an imaging quality inspection point of view; if the imaging quality inspection point of view characterizes that the liquid crystal imaging image to be analyzed is a quality inspection defect point of view, determining the optical image data to be analyzed as substandard optical image data.

9. An optical imaging data analysis system, characterized in that: include: Memory, used to store program instructions and data; A processor, coupled to a memory, and configured to execute instructions in the memory to implement the method according to any one of claims 1 to 8.

10. A computer storage medium, characterized in that: Contains instructions, which, when executed on a processor, implement the method according to any one of claims 1 to 8.

Images