Method for identifying camera module based on artificial intelligence

Abstract

The invention provides a method for identifying a camera module based on artificial intelligence. The method comprises the following steps: acquiring an appearance image of the camera module; acquiring an image acquisition mode of the camera module; obtaining a target image collected by a camera module, and extracting characteristic attributes of the target image; based on the appearance image, analyzing a first classification of the camera module; based on the image acquisition mode, analyzing a second classification of the camera module; based on the characteristic attributes, analyzing a third classification of the camera module; and performing fusion analysis on the first classification, the second classification and the third classification based on an artificial intelligence model, and identifying to obtain a target classification of the camera module. According to the invention, multiple key factors such as the appearance of the camera module, the image acquisition mode and the characteristic attribute of the target image are comprehensively considered, and accurate identification and classification of the camera module are realized through the fusion analysis technology.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method for identifying a camera module based on artificial intelligence. Background Art

[0002] Traditional camera module recognition technology often relies on manual experience or a single feature analysis method. For example, judging the type of camera module only by a cursory observation of the appearance is not only inaccurate, but also unable to gain an in-depth understanding of its internal performance characteristics and image acquisition capabilities. The analysis of image acquisition methods is also relatively simple, usually focusing only on some basic parameter settings, while ignoring its adaptability and stability in different environments and application scenarios.

[0003] In terms of target image processing, traditional methods simply analyze basic attributes of the image, such as clarity and color, but lack a comprehensive and in-depth exploration of the image’s characteristic attributes, and are unable to accurately infer the performance and characteristics of the camera module based on image features.

[0004] In addition, existing recognition methods have shown obvious limitations in the face of the increasingly diverse and complex camera module market. Camera modules produced by different manufacturers have different characteristics and parameters, and traditional methods are difficult to accurately classify and recognize various types of camera modules. Summary of the invention

[0005] The main purpose of the present invention is to provide a method for identifying a camera module based on artificial intelligence, aiming to overcome the current defect that the camera module cannot be accurately identified.

[0006] To achieve the above object, the present invention provides a method for identifying a camera module based on artificial intelligence, comprising the following steps:

[0007] Acquire the appearance image of the camera module; obtain the image acquisition method of the camera module; obtain the target image acquired by the camera module, and extract the characteristic attributes of the target image;

[0008] Based on the appearance image, analyzing the first classification of the camera module; based on the image acquisition method, analyzing the second classification of the camera module; based on the characteristic attributes, analyzing the third classification of the camera module;

[0009] The first classification, the second classification and the third classification are fused and analyzed based on the artificial intelligence model to identify the target classification of the camera module.

[0010] Furthermore, the image acquisition method of the camera module includes:

[0011] Detecting the hardware interface information of the camera module, comparing it with preset standard interface parameters, determining the acquisition mode category corresponding to the connection interface type, and obtaining interface-related acquisition mode information;

[0012] Analyze the signal transmission mode of the camera module in the working state, determine the relevant acquisition mode characteristics according to the signal transmission mode, and obtain signal mode acquisition mode information;

[0013] Parse the driver of the camera module, extract the driver features including the configuration parameters and instruction set of image acquisition, and determine the driver feature associated acquisition mode information according to the preset correspondence library between the driver feature and the acquisition mode;

[0014] The image acquisition method of the camera module is obtained by combining the interface-related acquisition method information, the signal mode acquisition method information and the drive feature-related acquisition method information and applying a preset fusion algorithm.

[0015] Further, the extracting characteristic attributes of the target image includes:

[0016] Performing color space conversion processing on the target image, converting it from RGB color space to HSV color space, and calculating the mean, variance, and peak value of each channel in the converted color space to obtain color feature attributes;

[0017] Using an edge detection algorithm to extract edges of the target image, calculate the length, density, and curvature of the edges, and obtain edge feature attributes;

[0018] Using a texture analysis algorithm to analyze the texture of the target image, extracting the roughness, contrast, and orientation of the texture, and obtaining texture feature attributes;

[0019] Divide the target image into multiple sub-blocks, calculate the gray level co-occurrence matrix of each sub-block, extract energy, entropy, and correlation features from them, and obtain the spatial distribution feature attributes of the target image by combining the features of each sub-block;

[0020] The color feature attributes, edge feature attributes, texture feature attributes and space distribution feature attributes are fused to obtain characteristic attributes of the target image.

[0021] Further, the analyzing the first classification of the camera module based on the appearance image; analyzing the second classification of the camera module based on the image acquisition method; and analyzing the third classification of the camera module based on the characteristic attributes include:

[0022] Matching the appearance image with appearance images of various types of camera modules stored in a database to obtain the probability that the camera module is of various types as a first classification;

[0023] Matching the image acquisition mode with the image acquisition modes of various types of camera modules stored in the database to obtain the probability that the camera module is of various types as the second classification;

[0024] The characteristic attributes are matched with the characteristic attributes of each type of camera module stored in the database to obtain the probability that the camera module is of each type as the third classification.

[0025] Furthermore, the first classification, the second classification and the third classification are fused and analyzed based on the artificial intelligence model to identify the target classification of the camera module, including:

[0026] Constructing a multidimensional feature space, mapping the first category, the second category, and the third category to different dimensional axes of the multidimensional feature space to form a feature vector point;

[0027] Obtaining an artificial intelligence model; wherein the artificial intelligence model is an autoencoder based on deep learning, which pre-learns the distribution of feature vector points of a known type of camera module in a multidimensional feature space to obtain potential distribution patterns and laws of the feature space;

[0028] Inputting the feature vector points of the camera module into the artificial intelligence model, and calculating the reconstruction error of the feature vector points in the feature space based on the autoencoder;

[0029] Determining whether the reconstruction error is within a threshold range;

[0030] If the reconstruction error is within the threshold range, the closest known type is found as the target classification based on the feature mapping relationship inside the autoencoder and the similarity measurement with the known type;

[0031] If the reconstruction error exceeds the threshold range, it is judged as a new type or an abnormal type, and further manual intervention mechanism is triggered.

[0032] Furthermore, after the first classification, the second classification and the third classification are fused and analyzed based on the artificial intelligence model and the target classification of the camera module is identified, the method further includes:

[0033] Encoding the target classification to obtain a first code;

[0034] Performing hash operations on the appearance image, the image acquisition method, and the characteristic attributes to obtain corresponding first hash values, second hash values, and third hash values;

[0035] Reconstructing the first hash value, the second hash value, and the third hash value to generate a hash value matrix;

[0036] The hash value matrix is ​​screened based on the first code to generate an identification code for identifying images subsequently captured by the camera module.

[0037] The present invention also provides a device for identifying a camera module based on artificial intelligence, comprising:

[0038] An acquisition unit is used to acquire an appearance image of the camera module; acquire an image acquisition method of the camera module; acquire a target image acquired by the camera module, and extract characteristic attributes of the target image;

[0039] An analysis unit, configured to analyze a first classification of the camera module based on the appearance image; analyze a second classification of the camera module based on the image acquisition method; and analyze a third classification of the camera module based on the characteristic attribute;

[0040] An identification unit is used to perform a fusion analysis on the first classification, the second classification and the third classification based on an artificial intelligence model to identify the target classification of the camera module.

[0041] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0042] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0043] The method for identifying a camera module based on artificial intelligence provided by the present invention includes: collecting an appearance image of a camera module; obtaining an image acquisition method of the camera module; obtaining a target image collected by the camera module, and extracting characteristic attributes of the target image; analyzing a first classification of the camera module based on the appearance image; analyzing a second classification of the camera module based on the image acquisition method; analyzing a third classification of the camera module based on the characteristic attributes; and performing a fusion analysis on the first classification, the second classification, and the third classification based on an artificial intelligence model to identify and obtain the target classification of the camera module. In the present invention, multiple key factors including the appearance of the camera module, the image acquisition method, and the characteristic attributes of the target image are comprehensively considered, and accurate identification and classification of the camera module are achieved through fusion analysis technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the steps of a method for identifying a camera module based on artificial intelligence in one embodiment of the present invention;

[0045] Figure 2 It is a structural block diagram of a device for recognizing a camera module based on artificial intelligence in one embodiment of the present invention;

[0046] Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0047] The implementation, functional features and advantages of the present invention will be further described in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0049] Reference Figure 1 In one embodiment of the present invention, a method for identifying a camera module based on artificial intelligence is provided, comprising the following steps:

[0050] Step S1, collecting an appearance image of the camera module; obtaining an image collection method of the camera module; obtaining a target image collected by the camera module, and extracting characteristic attributes of the target image;

[0051] Step S2, analyzing the first classification of the camera module based on the appearance image; analyzing the second classification of the camera module based on the image acquisition method; and analyzing the third classification of the camera module based on the characteristic attributes;

[0052] Step S3, performing a fusion analysis on the first classification, the second classification, and the third classification based on an artificial intelligence model to identify the target classification of the camera module.

[0053] Traditional methods usually rely only on appearance observation or simple image parameter analysis, and cannot fully and accurately identify the type of camera module. In this embodiment, by integrating artificial intelligence technology and multi-dimensional analysis, misjudgment and missed judgment are reduced, and a large amount of camera module data can be quickly processed.

[0054] As described in step S1 above, first, use a high-resolution image acquisition device to shoot the camera module from multiple angles to ensure that its appearance features, including shape, size, surface texture, etc., can be fully acquired. Perform preliminary preprocessing on the acquired images, such as removing noise, adjusting brightness and contrast, etc., to improve image quality and provide a clear data source for subsequent analysis.

[0055] At the same time, analyze the hardware configuration and driver of the camera module to understand the basic parameter settings of its image acquisition, such as resolution, frame rate, exposure time, etc. By communicating with the camera module, you can monitor its data transmission mode and protocol in actual work, including data format, transmission rate, etc., to fully understand the characteristics of its image acquisition method.

[0056] Furthermore, the actual images collected by the camera module are deeply analyzed, and image processing algorithms are used to extract various characteristic attributes of the image. For example, the color characteristics of the image are analyzed, including color distribution, color contrast, color saturation, etc.; the texture characteristics of the image are analyzed, such as the roughness, direction, and periodicity of the texture; the edge characteristics of the image are detected, including the clarity, quantity, and shape of the edge; and the spatial distribution characteristics of the image are analyzed, such as the positional relationship and distribution density of the objects in the image.

[0057] As described in step S2 above, the collected camera module appearance image is compared with the pre-established appearance feature database using image recognition and pattern matching technology. According to the similarity of appearance features, the camera modules are divided into different appearance categories, such as circular camera modules, square camera modules, special shape camera modules, etc. At the same time, it can be further subdivided, such as metal shell camera modules, plastic shell camera modules, etc. according to different surface materials. At the same time, the camera modules are classified according to the acquired image acquisition method parameters and features. For example, according to the frame rate, it is divided into high frame rate camera modules, medium frame rate camera modules and low frame rate camera modules; according to the resolution, it is divided into high-definition camera modules, standard definition camera modules, etc.; it can also be divided into different functional categories according to the particularity of the acquisition mode, such as whether it supports panoramic shooting, dynamic tracking and other functions.

[0058] At the same time, based on the extracted target image characteristics, the camera modules are classified using data analysis and machine learning algorithms. For example, according to the color characteristics, they can be divided into color-rich camera modules, single-color camera modules, etc.; according to the texture characteristics, they can be divided into texture-clear camera modules, texture-fuzzy camera modules, etc.; according to the edge characteristics and spatial distribution characteristics, corresponding classifications can also be made, such as sharp-edge camera modules, uniformly distributed object camera modules, etc.

[0059] As described in step S3 above, an artificial intelligence model capable of integrating multi-dimensional information is constructed in advance based on deep learning algorithms, such as convolutional neural networks, recurrent neural networks, etc. The model is trained and optimized with a large amount of training data so that it can accurately understand and process the results of the first classification, the second classification, and the third classification, and can perform comprehensive analysis and judgment based on this information.

[0060] In this embodiment, the results of the first classification, the second classification, and the third classification are used as input data and input into the artificial intelligence model. The model performs fusion analysis on these classification results through calculation and logical processing. For example, different weights are given to each classification result by weighted fusion, and adjusted according to its importance to the final target classification. At the same time, feature extraction and pattern recognition techniques can also be used to dig out potential associations and features between classification results. After fusion analysis, the artificial intelligence model outputs the final target classification result. The above results are a comprehensive evaluation and classification of camera modules, including their type information. For example, the model can output that the camera module belongs to a specific classification category such as a high-definition camera module for smartphones, a panoramic camera module for security monitoring, and a high and low temperature resistant camera module for industrial automation.

[0061] In this embodiment, the three key aspects of the camera module's appearance, image acquisition method, and target image characteristic attributes are comprehensively considered, which fully reflects the characteristics of the camera module and improves the accuracy and reliability of recognition. By using deep learning artificial intelligence technology to integrate and analyze multi-dimensional data, it is possible to automatically learn and mine the potential relationships and features between data, and achieve more intelligent recognition and classification. Compared with traditional methods, it can quickly process a large amount of camera module data and provide more accurate classification results, providing strong support for the production, application, and management of camera modules.

[0062] In one embodiment, the method of acquiring an image from a camera module includes:

[0063] Detecting the hardware interface information of the camera module, comparing it with preset standard interface parameters, determining the acquisition mode category corresponding to the connection interface type, and obtaining interface-related acquisition mode information;

[0064] Analyze the signal transmission mode of the camera module in the working state, determine the relevant acquisition mode characteristics according to the signal transmission mode, and obtain signal mode acquisition mode information;

[0065] Parse the driver of the camera module, extract the driver features including the configuration parameters and instruction set of image acquisition, and determine the driver feature associated acquisition mode information according to the preset correspondence library between the driver feature and the acquisition mode;

[0066] The image acquisition method of the camera module is obtained by combining the interface-related acquisition method information, the signal mode acquisition method information and the drive feature-related acquisition method information and applying a preset fusion algorithm.

[0067] In this embodiment, the conventional method of judging the camera module image acquisition mode is often based on a single factor, such as relying only on driver information or simple signal observation, resulting in inaccurate and incomplete judgment. This solution aims to achieve accurate acquisition mode judgment by integrating multiple factors.

[0068] Specifically, first use hardware detection equipment or software tools to comprehensively scan and analyze the physical interface of the camera module. These detection tools can read detailed information such as the electrical characteristics and physical structure of the interface. For example, for a common USB interface camera module, the detection equipment can identify key parameters such as the version of the interface (such as USB 2.0 or USB 3.0), the power supply capacity of the interface, and the data transmission rate.

[0069] A database containing standard parameters of various common camera module hardware interfaces is pre-established. This database covers the detailed specifications, characteristics and corresponding acquisition method category information of different interface types. The detected camera module hardware interface information is compared with the standard parameters in the database one by one. For example, if the detected interface data transmission rate is within a certain range and the physical structure of the interface meets the characteristics of a certain standard interface type, its connection interface type can be preliminarily determined. According to the comparison results, the acquisition method category to which the connection interface type of the camera module belongs is determined. For example, some specific types of interfaces are mainly used for continuous image acquisition, while other interfaces are more suitable for specific modes of image acquisition (such as high-speed continuous shooting mode). The above information is organized into interface-related acquisition method information to provide basic data for subsequent comprehensive analysis.

[0070] Furthermore, signal monitoring equipment and software are used to monitor the signal transmission of the camera module in real time when it is in working state. These devices can capture various characteristics of the signal, including the frequency, amplitude, phase, etc. of the signal. For example, for a camera module that adopts wireless transmission, the signal monitoring device can monitor the transmission frequency of its wireless signal, the change pattern of the signal strength, etc. According to the monitored signal transmission mode, its characteristics are analyzed and the relevant acquisition mode characteristics are determined. For example, if the frequency of the signal is stable and within a specific range, it means that the camera module adopts an image acquisition method with a fixed frame rate; if the amplitude of the signal changes greatly and has a certain regularity, it means that it is performing some kind of dynamic image acquisition (such as signal changes during autofocus). The acquisition mode characteristics obtained by the analysis are sorted and summarized to form signal mode acquisition mode information. The above information describes the image acquisition mode characteristics reflected by the camera module in terms of signal transmission.

[0071] Furthermore, a driver parsing tool is used to conduct an in-depth analysis of the driver of the camera module. The above tool can read various configuration parameters, instruction sets and key information related to image acquisition in the driver. For example, key information such as parameters for image resolution setting, instruction sets for frame rate control, and image encoding methods in the driver are extracted. A database containing the correspondence between various driver features and corresponding acquisition methods is pre-built. This database is established through the analysis of driver programs of a large number of different types of camera modules and the study of actual acquisition methods. For example, a certain combination of driver configuration parameters corresponds to a specific image acquisition mode (such as night shooting mode). According to the extracted driver features, search and match are performed in the correspondence library. Determine the acquisition method information associated with the driver feature. For example, if the extracted driver parameters show support for high-resolution image acquisition and fast autofocus functions, it can be inferred that the corresponding acquisition method is a mode suitable for high-quality image capture.

[0072] Finally, a fusion algorithm is pre-configured that comprehensively considers the interface-related acquisition mode information, the signal mode acquisition mode information, and the drive feature-related acquisition mode information. The algorithm can be combined with a variety of technical means such as weighted average, fuzzy logic judgment, and neural network. For example, different weights are assigned to each source of information according to its reliability and importance. The acquisition mode information from the three sources is input into the fusion algorithm for comprehensive analysis. The algorithm fuses this information according to preset rules and logic. For example, if the interface-related acquisition mode information shows that the camera module is suitable for a certain continuous image acquisition mode, and the signal mode acquisition mode information and the drive feature-related acquisition mode information also have corresponding features to support this conclusion, then the fusion algorithm can more certainly determine that the image acquisition mode of the camera module belongs to this continuous image acquisition mode. After being processed by the fusion algorithm, the image acquisition mode of the camera module that is finally determined is output. This result is an accurate judgment based on information from multiple aspects, including detailed information such as the type, characteristics, and parameters of the acquisition mode.

[0073] In this embodiment, information from three key aspects, namely, hardware interface, signal transmission mode and driver, is integrated to determine the image acquisition method, thereby overcoming the limitation of a single information source and improving the accuracy and reliability of judgment.

[0074] In one embodiment, extracting the characteristic attributes of the target image includes:

[0075] Performing color space conversion processing on the target image, converting it from RGB color space to HSV color space, and calculating the mean, variance, and peak value of each channel in the converted color space to obtain color feature attributes;

[0076] Using an edge detection algorithm to extract edges of the target image, calculate the length, density, and curvature of the edges, and obtain edge feature attributes;

[0077] Using a texture analysis algorithm to analyze the texture of the target image, extracting the roughness, contrast, and orientation of the texture, and obtaining texture feature attributes;

[0078] Divide the target image into multiple sub-blocks, calculate the gray level co-occurrence matrix of each sub-block, extract energy, entropy, and correlation features from them, and obtain the spatial distribution feature attributes of the target image by combining the features of each sub-block;

[0079] The color feature attributes, edge feature attributes, texture feature attributes and space distribution feature attributes are fused to obtain characteristic attributes of the target image.

[0080] In this embodiment, first, the target image is converted from the common RGB color space to the HS color space. This is because the HSV space is more in line with the way humans perceive colors and can better separate the color hue, saturation and brightness information. This conversion process is implemented using an image processing algorithm to ensure the accuracy and efficiency of the conversion. In the converted HSV color space, statistical analysis is performed on the three channels of H (hue), S (saturation), and V (brightness). The mean of each channel is calculated to reflect the average level of the color value of the channel; the variance reflects the discrete degree of the color value, that is, the diversity of the color; the peak value represents the color value and its intensity that appear most frequently in the channel. Through the above statistical values, the characteristic distribution of the target image in terms of color can be fully described.

[0081] Furthermore, edge detection algorithms such as the Canny algorithm and the Sobel algorithm are used to detect the edges of the target image. First, the length of the edge is calculated, that is, the total length of all edge segments in the image, which reflects the complexity and size of the object boundary. Density represents the number of edge segments per unit area, which can be used to measure the detail richness and texture complexity of the object in the image. Curvature describes the degree of curvature of the edge curve, which is important for identifying objects with specific shape features such as circles and arcs.

[0082] Furthermore, texture analysis algorithms, such as gray-level co-occurrence matrix method and wavelet transform method, are used to conduct in-depth analysis of the texture of the target image. The above algorithms can capture the gray-level relationship and spatial distribution law between pixels in the image, thereby revealing the texture characteristics of the image.

[0083] Among them, roughness indicates the roughness or smoothness of the texture, which is determined by calculating the gradient of the pixel grayscale value or counting the differences between pixels. Contrast reflects the degree of difference in grayscale values ​​in the texture, that is, the contrast intensity between bright and dark areas. Directionality describes the main direction of the texture or the trend of the texture lines, which can be obtained by calculating the correlation of pixel grayscale values ​​or using a directional filter.

[0084] Furthermore, the target image is divided into multiple sub-blocks, and the method of uniform block division or adaptive block division according to the image content can be adopted. Uniform block division is simple and intuitive, and is easy to calculate and process; adaptive block division can be flexibly divided according to the distribution and structure of objects in the image, and better adapt to different types of images. For each sub-block, its gray level co-occurrence matrix is ​​calculated. The gray level co-occurrence matrix reflects the joint probability distribution of gray values ​​in the image at a certain spatial distance and direction. Energy, entropy, correlation and other features are extracted from the gray level co-occurrence matrix. Energy represents the sum of squares of matrix element values, reflecting the uniformity and regularity of image texture; entropy represents the complexity and randomness of image texture; correlation describes the similarity of matrix elements in row or column direction, reflecting the linear characteristics of image texture. Then, the features of each sub-block are integrated to obtain the spatial distribution feature attributes of the target image. Simple averaging, weighted averaging or more complex fusion algorithms can be used to make reasonable integration according to factors such as the position and importance of sub-blocks.

[0085] Finally, the color feature attributes, edge feature attributes, texture feature attributes, and spatial distribution feature attributes are fused through the fusion algorithm. The feature vector splicing method can be used to combine the feature vectors into a comprehensive feature vector in a certain order; or the weighted fusion method can be used to assign different weights to each feature according to its importance, and then perform weighted summation. The fusion algorithm can be implemented by programming, and the extracted feature attribute data is processed according to the fusion strategy. In the fusion process, the dimensional consistency of the features and the compatibility of the data types need to be considered to ensure the accuracy and effectiveness of the fusion results. After the fusion process, the comprehensive feature attribute description of the target image is obtained. This feature attribute contains key information of the image in terms of color, edge, texture, spatial distribution, etc., and can fully and accurately reflect the characteristics of the target image.

[0086] In this embodiment, a feature extraction method based on four key dimensions, namely color, edge, texture and spatial distribution, is integrated to comprehensively describe the characteristics of the target image and avoid the limitation of single feature analysis.

[0087] In one embodiment, the analyzing the first classification of the camera module based on the appearance image; analyzing the second classification of the camera module based on the image acquisition method; and analyzing the third classification of the camera module based on the characteristic attribute include:

[0088] Matching the appearance image with appearance images of various types of camera modules stored in a database to obtain the probability that the camera module is of various types as a first classification;

[0089] Matching the image acquisition mode with the image acquisition modes of various types of camera modules stored in the database to obtain the probability that the camera module is of various types as the second classification;

[0090] The characteristic attributes are matched with the characteristic attributes of each type of camera module stored in the database to obtain the probability that the camera module is of each type as the third classification.

[0091] In this embodiment, the traditional camera module classification method often relies on manual experience or single feature judgment, which has low accuracy and efficiency. This solution aims to use advanced technical means to achieve automatic and accurate classification.

[0092] First, establish a database containing a large number of different types of camera module appearance images. This database should cover various common and special camera module appearance styles, including images with different shapes, sizes, materials, surface textures and other features. Each appearance image in the database is annotated and classified in detail to clarify the type of camera module it belongs to. The image matching algorithm can be a feature point-based matching algorithm or a deep learning-based image recognition algorithm. For the appearance image of the camera module to be classified, extract its key feature points or extract high-level feature representation through a neural network. Compare and match the above features with the appearance image features in the database one by one. According to the matching results, calculate the similarity between the appearance image of the camera module to be classified and the appearance images of each type in the database. Usually, similarity metrics such as Euclidean distance and cosine similarity can be used to convert the similarity into a probability value. The probability distribution of the camera module belonging to each type is obtained, and the types with the highest probability values ​​are taken as the first classification results.

[0093] Secondly, a database containing the image acquisition method features of various types of camera modules is pre-built. Record in detail the resolution settings, frame rate ranges, acquisition modes (such as continuous acquisition, intermittent acquisition, triggered acquisition, etc.) and signal transmission parameters of different types of camera modules. Perform detailed parameter extraction and feature analysis on the image acquisition methods of the classified camera modules. Match and compare the above features with the image acquisition method features in the database. For example, compare whether the frame rate is within the typical frame rate range of a certain type, whether the acquisition mode matches a certain type, etc. Based on the matching results, calculate the degree of similarity and convert it into a probability value. Determine the probability distribution of the camera module belonging to each type in terms of image acquisition method, thereby obtaining the result of the second classification.

[0094] Furthermore, a database containing characteristic attribute data of various camera modules is pre-built. The above-mentioned characteristic attributes include, but are not limited to, color characteristic attributes (such as color mean, variance, etc.), edge characteristic attributes (such as edge length, curvature, etc.), texture characteristic attributes (such as roughness, contrast, etc.) and spatial distribution characteristic attributes (such as sub-block feature comprehensive value, etc.) of the target image. The characteristic attributes of each type of camera module are statistically analyzed and their typical feature models are established. The characteristic attributes extracted from the classified camera modules are quantified and described in detail. These characteristic attributes are matched and compared with the characteristic attributes of various camera modules in the database. The degree of similarity with each type is evaluated through the classification algorithm or similarity measurement method in machine learning. The probability value is calculated according to the degree of similarity to obtain the probability distribution of the camera module belonging to each type in terms of characteristic attributes. This is used as the result of the third classification.

[0095] In this embodiment, the classification is performed by combining the three key aspects of appearance image, image acquisition method and characteristic attributes, comprehensively considering various characteristics of the camera module, and improving the accuracy and reliability of the classification.

[0096] In one embodiment, the fusion analysis of the first classification, the second classification, and the third classification based on the artificial intelligence model to identify the target classification of the camera module includes:

[0097] Constructing a multidimensional feature space, mapping the first category, the second category, and the third category to different dimensional axes of the multidimensional feature space to form a feature vector point;

[0098] Obtaining an artificial intelligence model; wherein the artificial intelligence model is an autoencoder based on deep learning, which pre-learns the distribution of feature vector points of a known type of camera module in a multidimensional feature space to obtain potential distribution patterns and laws of the feature space;

[0099] Inputting the feature vector points of the camera module into the artificial intelligence model, and calculating the reconstruction error of the feature vector points in the feature space based on the autoencoder;

[0100] Determining whether the reconstruction error is within a threshold range;

[0101] If the reconstruction error is within the threshold range, the closest known type is found as the target classification based on the feature mapping relationship inside the autoencoder and the similarity measurement with the known type;

[0102] If the reconstruction error exceeds the threshold range, it is judged as a new type or an abnormal type, and further manual intervention mechanism is triggered.

[0103] In this embodiment, first, according to the characteristics and properties of the first classification, the second classification and the third classification, the dimensional structure of the multidimensional feature space is determined. For example, the first classification can be corresponded to a dimensional axis, representing the appearance feature dimension of the camera module; the second classification can be corresponded to another dimensional axis, representing the image acquisition method feature dimension; the third classification can be corresponded to the third dimensional axis, representing the characteristic attribute feature dimension. The scale and value range on each dimensional axis are set according to the specific feature value range of the corresponding classification. The result value of the first classification is mapped to the corresponding dimensional axis of the multidimensional feature space according to certain rules to form a coordinate point on the dimension. For example, if the first classification is based on the shape classification of the appearance image, the circle corresponds to the value 1 and the square corresponds to the value 2, then when the first classification result of a camera module is a circle, it is marked as the position of the value 1 on the appearance feature dimension axis. Similarly, the results of the second classification and the third classification are respectively mapped to the corresponding dimensional axes to form a complete representation of the feature vector point in the multidimensional feature space.

[0104] Furthermore, an autoencoder based on deep learning is used as the core model. The autoencoder is an unsupervised learning model that can automatically learn the potential feature representation of data. According to the characteristics of the camera module data and the requirements of the classification task, a suitable autoencoder structure is designed and constructed. For example, an autoencoder with a multi-layer neural network structure can be selected, including an input layer, a hidden layer, and an output layer. The number of nodes in the input layer and the output layer is determined according to the dimension of the multidimensional feature space, and the structure and number of nodes in the hidden layer are determined through experiments and optimization to achieve effective extraction and compression of data features. The model is trained using feature vector point data of a large number of known types of camera modules in the multidimensional feature space. These data are input into the autoencoder, and the parameters of the autoencoder are continuously adjusted through the back propagation algorithm, so that the autoencoder can learn the potential distribution pattern and law of these data in the feature space. During the training process, a variety of optimization algorithms and techniques, such as stochastic gradient descent, adaptive learning rate adjustment, etc., can be used to improve training efficiency and model performance. At the same time, regularization methods, such as L1 regularization, L2 regularization, etc., can also be used to prevent model overfitting.

[0105] The feature vector points of the camera module to be identified are input into the trained autoencoder. The autoencoder will try to reconstruct the input feature vector points according to the feature mapping relationship it has learned, that is, it will be encoded from the input layer through the hidden layer and then decoded and restored through the output layer. Pre-select a suitable error calculation indicator, such as mean square error (MSE) or mean absolute error (MAE). Calculate the error value between the input feature vector point and the feature vector point reconstructed and output by the autoencoder. For example, for the mean square error, calculate the square of the difference in each dimension, then sum and average to obtain the final reconstruction error value. Through the reconstruction error analysis and experiments of a large number of known types of camera modules, a reasonable reconstruction error threshold range is determined. This threshold range should be able to distinguish between normal known types and new types or abnormal types. In one embodiment, it can also be adjusted and optimized according to different application scenarios and requirements for classification accuracy.

[0106] If the reconstruction error is within the threshold range, it means that the characteristics of the camera module are similar to the characteristic distribution pattern of the known types. At this time, according to the feature mapping relationship inside the autoencoder and the similarity measurement method with the known types, find the closest known type in the feature space as the target classification. There are many methods for similarity measurement, such as Euclidean distance, cosine similarity, etc. Calculate the similarity measurement value between the feature vector point of the camera module to be identified and the feature vector point of each known type, and select the known type with the smallest measurement value (i.e. the most similar) as the target classification result. If the reconstruction error exceeds the threshold range, it means that the characteristics of the camera module are quite different from the known types, which may be a new type or an abnormal situation. At this time, the system determines it as a new type or an abnormal type, and triggers further manual intervention mechanism.

[0107] When it is determined to be a new type or an abnormal type, the system will issue a prompt message to notify relevant personnel to intervene manually. At the same time, it automatically collects and organizes detailed information related to the camera module, including its original data (such as appearance image, image acquisition method parameters, target image characteristic attributes, etc.), the specific value of the reconstruction error and the location information of the feature vector point in the multidimensional feature space, etc. Professionals conduct a detailed analysis and judgment of the camera module based on the information provided by the system. Professional knowledge and experience can be used in combination with other auxiliary information (such as the manufacturer of the camera module, application scenarios, etc.) to determine whether it is a new type. If it is a new type, it is manually classified and labeled, and its relevant information is added to the database for subsequent model learning and classification. If it is an abnormal type, further analyze the cause of the abnormality, such as data acquisition errors, equipment failures, etc., and take corresponding treatment measures.

[0108] In one embodiment, after the first classification, the second classification, and the third classification are fused and analyzed based on the artificial intelligence model to identify the target classification of the camera module, the following steps are included:

[0109] Encoding the target classification to obtain a first code;

[0110] Performing hash operations on the appearance image, the image acquisition method, and the characteristic attributes to obtain corresponding first hash values, second hash values, and third hash values;

[0111] Reconstructing the first hash value, the second hash value, and the third hash value to generate a hash value matrix;

[0112] The hash value matrix is ​​screened based on the first code to generate an identification code for identifying images subsequently captured by the camera module.

[0113] In this embodiment, after completing the target classification and recognition of the camera module, the classification results need to be encoded and stored for subsequent data query, management and application. Traditional methods lack efficient encoding and data association mechanisms, resulting in chaotic data management and difficult retrieval. This solution solves this problem through encoding and hashing. For the large amount of image data collected by the camera module, a reliable identification method is needed to distinguish between images collected by different modules and at different stages, so as to facilitate the tracking, analysis and organization of image data. Existing identification methods are not flexible enough or difficult to effectively associate with the classification information of the camera module. This solution aims to establish an image identification system based on classification information.

[0114] First, a set of coding rules for target classification is formulated in advance. The coding rules can be designed according to the classification hierarchy, type characteristics and application requirements. For example, digital coding, letter coding or mixed coding can be used. If the classification of camera modules includes different application fields (such as security, medical, transportation, etc.) and different performance levels (such as high, medium and low), a coding system containing field codes and performance level codes can be designed. The coding rules should be unique and extensible so that they can adapt to new classifications and detailed classifications that will appear in the future. According to the designed coding rules, the target classification of the identified camera module is converted into the corresponding code, that is, the first code. The above code can be a short string or a digital sequence for easy storage and processing. For example, if a camera module is identified as a high-performance type in the security field, a code similar to "A18C977" can be generated according to the coding rules.

[0115] Then, a hash algorithm, such as MD5, SHA-256, etc., is selected to perform hash calculation on the appearance image of the camera module. The hash algorithm converts the appearance image data into a hash value of a fixed length, namely the first hash value. This hash value is unique, and even if there is a slight change in the appearance image, a different hash value will be generated. Therefore, the differences between different appearance images can be quickly identified and compared, and the appearance images can be stored and retrieved in the form of hash values, saving storage space and improving processing efficiency. Similarly, the selected hash algorithm is used to perform hash calculation on the image acquisition method related data of the camera module. The above data includes information such as resolution settings, frame rate parameters, and acquisition mode. This information is converted into a second hash value for uniquely identifying the image acquisition method characteristics of the camera module. Similarly, the characteristic attribute data of the camera module is hashed, including the color feature attributes, edge feature attributes, texture feature attributes, and spatial distribution feature attributes of the target image. The above complex data is converted into a concise third hash value through a hash algorithm to facilitate subsequent data processing and association.

[0116] The above matrix can be two-dimensional or multi-dimensional, and the specific dimensions and sizes can be adjusted according to the complexity of the camera module and the amount of data. For example, a two-dimensional matrix can be designed, in which the first row is used to store fragments of the first hash value, the second row is used to store fragments of the second hash value, and the third row is used to store fragments of the third hash value. Each hash value can be divided into multiple fragments according to certain rules to facilitate reasonable distribution and storage in the matrix. The first hash value, the second hash value, and the third hash value are reconstructed according to the designed matrix structure. For example, the first hash value can be divided into several sub-fragments of equal length and filled into the corresponding positions of the first row of the matrix in turn; the second hash value and the third hash value are also filled into the second and third rows of the matrix in the same way. During the reconstruction process, some check information or padding characters can also be added to ensure the integrity and consistency of the matrix.

[0117] The association relationship between the first code and the hash value matrix is ​​established in advance. This can be achieved by adding a pointer, index or other associated fields in the data structure. When it is necessary to identify the image subsequently collected by the camera module, the corresponding hash value matrix can be quickly found according to the first code.

[0118] Pre-configure a screening rule based on the first code to extract key information from the hash value matrix to generate an identification code. The screening rule can be customized according to application requirements and data characteristics. For example, hash value fragments at certain specific positions in the matrix can be selected for combination according to the specific bit value of the first code; or the hash values ​​in the matrix can be weighted according to the classification information of the code to obtain a comprehensive identification code. According to the screening rule, relevant information is extracted from the hash value matrix, and combined, converted or calculated to generate an identification code for identifying the subsequent image acquisition of the camera module. The identification code can be a combination of numbers, letters or symbols, which is unique and identifiable. For example, in one embodiment, if the first code represents a specific type of camera module, the information extracted from the hash value matrix can be combined and converted to generate an identification code similar to "CAM-001-A18C977", where "CAM-001" represents the serial number of the camera module and "A18C977" represents its target classification code.

[0119] In this embodiment, an efficient data management and association mechanism is established by encoding the target classification and associating it with the hash value matrix. The corresponding camera module data and image identification information can be easily and quickly found according to the code, which improves the efficiency of data retrieval and application. The method of hash value reconstruction generation matrix is ​​innovatively adopted to organically integrate different types of hash values ​​to form a unified data structure. This matrix structure can not only save storage space, but also facilitate data comparison, analysis and processing, providing a rich source of information for subsequent image identification.

[0120] In one embodiment, reconstructing the first hash value, the second hash value, and the third hash value to generate a hash value matrix includes:

[0121] Extracting the numeric characters in the first hash value and adding them to a matrix of three rows and three columns to obtain a first matrix;

[0122] Extracting the numeric characters in the second hash value and adding them to a matrix of three rows and three columns to obtain a second matrix;

[0123] Extracting English characters from the third hash value and adding them to a matrix of three rows and three columns to obtain a third matrix;

[0124] Adding the digital characters at corresponding positions in the first matrix and the second matrix to obtain a fourth matrix;

[0125] The third matrix is ​​multiplied by the fourth matrix to obtain the hash value matrix.

[0126] In this embodiment, for the first hash value, the digital characters therein are accurately identified and extracted. The above digital characters represent certain feature codes or feature values ​​related to the appearance image of the camera module. For example, it can be the size parameters of the appearance image after specific processing, the digital part in the color code, etc. According to the preset rules and order, the extracted digital characters are added to an initial blank matrix of three rows and three columns in sequence to form a first matrix. For example, each position of the matrix can be filled one by one in the order from left to right and from top to bottom.

[0127] The digital characters in the second hash value are extracted using a method similar to that of the first hash value, and the extracted digital characters are added to another blank matrix of three rows and three columns according to the same rule to obtain a second matrix.

[0128] For the third hash value, English characters are identified and extracted from the third hash value. The extracted English characters are added to a blank matrix of three rows and three columns according to a preset order and rule to form a third matrix. For example, the third matrix can be filled according to the importance of the characteristic attributes represented by the English characters or a preset classification order.

[0129] In order to further fuse the information in the first matrix and the second matrix, a simple and effective addition operation is designed. According to the corresponding positions of the matrices, the digital characters at the same positions in the first matrix and the second matrix are added. For example, if the number at a certain position in the first matrix is ​​"5" and the number at the corresponding position in the second matrix is ​​"3", then "8" will be obtained at the corresponding position in the fourth matrix. If the result after addition exceeds a certain numerical range (for example, it exceeds the representable range of digital characters or a reasonable range set according to actual application requirements), modulo operation or other appropriate processing methods can be used to ensure that the result is within a reasonable range and can accurately reflect the fusion of information.

[0130] All corresponding positions in the first matrix and the second matrix are added one by one, and a new three-row and three-column matrix is ​​obtained after completion, that is, the fourth matrix. This matrix integrates digital information related to the appearance image of the camera module and the image acquisition method.

[0131] Finally, the third matrix is ​​multiplied with the fourth matrix. For example, the multiplication process can be directly performed based on the matrix operation method, or a multiplication rule based on the combination of character encoding and digital operation can be defined. For each element at the corresponding position, the English characters in the third matrix are converted into digital values ​​according to the preset encoding conversion rules (for example, the corresponding digital values ​​are assigned according to the order of the letters in the alphabet), and then multiplied with the numbers at the corresponding positions in the fourth matrix.

[0132] After the multiplication operation, a new three-row and three-column matrix is ​​obtained, which is the final hash value matrix. The above matrix contains the fusion information from the first hash value, the second hash value and the third hash value, which not only reflects the fusion of digital features, but also contains the characteristic attribute information represented by English characters. It has a unique structure and information representation capability, and can be used as a comprehensive feature representation of the camera module for subsequent data analysis, identification generation, security management and other applications.

[0133] In this embodiment, by adding the first matrix and the second matrix and multiplying the matrix with the third matrix, a multi-level fusion of hash value information from different sources is achieved, forming a hash value matrix with rich connotations and unique structure, which provides a more comprehensive and accurate information basis for subsequent applications. The generated hash value matrix can be flexibly interpreted and processed according to different application requirements. For example, in image identification, a unique identification code can be generated based on the characters or numbers at specific positions in the matrix; in the field of data security, the complexity of the matrix can be used for encryption and decryption operations; in data analysis, the characteristics and patterns of the matrix can be analyzed to gain an in-depth understanding of the characteristics and performance of the camera module.

[0134] In one embodiment, the step of screening the hash value matrix based on the first code to generate an identification code includes:

[0135] Add the characters in the first code one by one in sequence into a matrix to generate a code matrix; wherein the number of rows and the number of columns of the code matrix are both smaller than the number of rows and the number of columns of the hash value matrix;

[0136] The encoding matrix is ​​superimposed on the upper left corner of the hash value matrix, and the number of superimposed positions of the encoding matrix and the hash value matrix that meet the preset condition is calculated as the target number; so the preset condition is that the types of characters at the corresponding positions are the same;

[0137] The encoding matrix is ​​translated sequentially, and when the target number reaches a maximum for the first time, an area in the current hash value matrix that overlaps with the encoding matrix is ​​obtained as a target area, and characters in the target area are combined to obtain a target character combination;

[0138] According to the rules, the target character combination is processed to obtain the identification code.

[0139] In this embodiment, the traditional identification code generation method often lacks a close association with the core classification information of the camera module (i.e., the first code), which makes it difficult for the identification code to directly reflect the key features and classification of the camera module. This solution realizes accurate identification generation based on the characteristics of the camera module itself by organically combining the first code with the hash value matrix. This solution can quickly generate highly distinguishable identification codes through an innovative matrix superposition and screening mechanism.

[0140] Specifically, characters are first extracted one by one from the first code. The first code is usually a code sequence with a specific meaning generated according to the target classification of the camera module, which contains character representations of key information such as the type and performance of the camera module. For example, A35CT682S734.

[0141] According to the preset arrangement rules, the extracted characters are arranged in a new matrix in sequence to generate a coding matrix. For example, the characters can be filled into a small matrix with fewer rows and columns than the hash value matrix in order from left to right and from top to bottom. The size of this small matrix should be designed to be able to interact with the hash value matrix effectively without excessively occupying computing resources and storage space.

[0142] Then, a corresponding data structure is allocated for each position of the encoding matrix to store character information and intermediate data and tag information required in subsequent calculations. For example, a data structure containing the character itself can be set for each position.

[0143] Secondly, accurately superimpose the generated coding matrix on the upper left corner of the hash value matrix. This is the initial superposition position, and the translation operation will be performed as needed later. At this point, each position of the coding matrix forms a corresponding relationship with the corresponding position of the hash value matrix. The preset condition is defined as the same type of characters at the corresponding positions. The character types here can be classified according to actual application requirements, for example, they can be divided into digital types, letter types (further subdivided into uppercase and lowercase letters), special symbol types, etc. Compare the types of corresponding characters at the superimposed positions of the coding matrix and the hash value matrix one by one. If the types are the same, add one to the match count of the position, and record the position information that meets the preset conditions. Count the number of all positions that meet the conditions as the target number.

[0144] According to the preset translation rules, the coding matrix is ​​translated on the hash value matrix in sequence. For example, it can be translated one unit position (such as the width or height of a matrix position) each time in the order from left to right and from top to bottom. During the translation process, the above-mentioned character type comparison and target quantity statistical steps are repeated continuously to update the value of the target quantity in real time. Continuously monitor the changes in the target quantity. When the target quantity reaches the maximum value for the first time, record the position of the coding matrix on the hash value matrix at this time. The area in the hash value matrix that overlaps with the coding matrix is ​​determined as the target area. This target area contains the portion of the hash value matrix that has the greatest match and correlation with the first code, and is the key information source area for generating the identification code.

[0145] All characters are extracted from the target area and combined in the order of their positions in the target area to form a target character combination. The target character combination is further processed according to pre-designed rules to obtain the final identification code. The processing rules may include but are not limited to the following methods:

[0146] Character encoding conversion: convert the characters in the target character combination, such as converting letters into corresponding numeric codes, or converting special symbols into specific character sequences according to mapping rules.

[0147] Application of encryption algorithm: Use an encryption algorithm to encrypt the target character combination to enhance the security and uniqueness of the identification code. For example, a simple Caesar cipher encryption method can be used to shift each character in the target character combination according to a certain offset.

[0148] Check digit addition: To ensure the accuracy and integrity of the identification code, a check digit can be added after the target character combination is processed. The check digit can be obtained by performing some calculation (such as hash calculation, parity calculation, etc.) on the processed characters and appended to the end of the identification code.

[0149] Length adjustment: The length of the generated identification code can be adjusted according to actual application requirements. Specific characters can be truncated or filled to make the length of the identification code meet specific standards or specifications.

[0150] In this embodiment, by converting the first code into a coding matrix and interacting with the hash value matrix, accurate screening and identification generation based on the classification information of the camera module is achieved. This combination method makes full use of the semantic information of the code and the structural information of the matrix, providing a rich data source and a unique processing method for the generation of identification codes. The method of translating the coding matrix is ​​adopted to search for the best matching area on the hash value matrix. This dynamic processing method improves the flexibility and adaptability of generating identification codes. It can automatically find the most representative target area according to different first codes and hash value matrix distributions, thereby generating a highly distinguishing identification code. The entire technical solution designs an efficient identification code generation process. From the generation of the coding matrix to the determination of the target area, and then to the processing of the final identification code, each step is carefully designed and optimized to ensure that the identification code can be generated quickly and accurately when processing a large amount of camera module data, thereby improving the efficiency of data processing and the performance of the system.

[0151] Reference Figure 2 In another embodiment of the present invention, a device for recognizing a camera module based on artificial intelligence is provided, comprising:

[0152] An acquisition unit is used to acquire an appearance image of the camera module; acquire an image acquisition method of the camera module; acquire a target image acquired by the camera module, and extract characteristic attributes of the target image;

[0153] An analysis unit, configured to analyze a first classification of the camera module based on the appearance image; analyze a second classification of the camera module based on the image acquisition method; and analyze a third classification of the camera module based on the characteristic attribute;

[0154] An identification unit is used to perform a fusion analysis on the first classification, the second classification and the third classification based on an artificial intelligence model to identify the target classification of the camera module.

[0155] In this embodiment, for the specific implementation of each unit in the above device embodiment, please refer to the above method embodiment, which will not be repeated here.

[0156] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0157] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0158] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0159] In summary, the method for identifying a camera module based on artificial intelligence provided in an embodiment of the present invention includes: acquiring an appearance image of a camera module; acquiring an image acquisition method of the camera module; acquiring a target image acquired by the camera module, and extracting characteristic attributes of the target image; analyzing a first classification of the camera module based on the appearance image; analyzing a second classification of the camera module based on the image acquisition method; analyzing a third classification of the camera module based on the characteristic attributes; and performing a fusion analysis of the first classification, the second classification, and the third classification based on an artificial intelligence model to identify the target classification of the camera module. In the present invention, multiple key factors including the appearance of the camera module, the image acquisition method, and the characteristic attributes of the target image are comprehensively considered, and accurate identification and classification of the camera module are achieved through fusion analysis technology.

[0160] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0161] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0162] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for identifying a camera module based on artificial intelligence, characterized in that: The following steps are involved: Collect appearance images of the camera module; Get the image acquisition method of the camera module; Acquire a target image captured by a camera module, and extract characteristic attributes of the target image; Analyzing a first classification of the camera module based on the appearance image; Based on the image acquisition method, analyzing the second classification of the camera module; based on the characteristic attribute, analyzing the third classification of the camera module; The first classification, the second classification and the third classification are fused and analyzed based on the artificial intelligence model to identify the target classification of the camera module.

2. The method for identifying camera modules based on artificial intelligence according to claim 1, characterized in that: The method of acquiring the image of the camera module includes: Detecting the hardware interface information of the camera module, comparing it with preset standard interface parameters, determining the acquisition mode category corresponding to the connection interface type, and obtaining interface-related acquisition mode information; Analyze the signal transmission mode of the camera module in the working state, determine the relevant acquisition mode characteristics according to the signal transmission mode, and obtain signal mode acquisition mode information; Parse the driver of the camera module, extract the driver features including the configuration parameters and instruction set of image acquisition, and determine the driver feature associated acquisition mode information according to the preset correspondence library between the driver feature and the acquisition mode; The image acquisition method of the camera module is obtained by combining the interface-related acquisition method information, the signal mode acquisition method information and the drive feature-related acquisition method information and applying a preset fusion algorithm.

3. The method for identifying camera modules based on artificial intelligence according to claim 1, characterized in that: The extracting characteristic attributes of the target image includes: Performing color space conversion processing on the target image, converting it from RGB color space to HSV color space, and calculating the mean, variance, and peak value of each channel in the converted color space to obtain color feature attributes; Using an edge detection algorithm to extract edges of the target image, calculate the length, density, and curvature of the edges, and obtain edge feature attributes; Using a texture analysis algorithm to analyze the texture of the target image, extracting the roughness, contrast, and orientation of the texture, and obtaining texture feature attributes; Divide the target image into multiple sub-blocks, calculate the gray level co-occurrence matrix of each sub-block, extract energy, entropy, and correlation features from them, and obtain the spatial distribution feature attributes of the target image by combining the features of each sub-block; The color feature attributes, edge feature attributes, texture feature attributes and space distribution feature attributes are fused to obtain characteristic attributes of the target image.

4. The method for identifying camera modules based on artificial intelligence according to claim 1, characterized in that: analyzing a first classification of the camera module based on the appearance image; Based on the image acquisition method, analyzing the second classification of the camera module; based on the characteristic attribute, analyzing the third classification of the camera module, including: Matching the appearance image with appearance images of various types of camera modules stored in a database to obtain the probability that the camera module is of various types as a first classification; Matching the image acquisition mode with the image acquisition modes of various types of camera modules stored in the database to obtain the probability that the camera module is of various types as the second classification; The characteristic attributes are matched with the characteristic attributes of each type of camera module stored in the database to obtain the probability that the camera module is of each type as the third classification.

5. The method for identifying camera modules based on artificial intelligence according to claim 1, characterized in that: The performing fusion analysis on the first classification, the second classification and the third classification based on the artificial intelligence model to identify the target classification of the camera module includes: Constructing a multidimensional feature space, mapping the first category, the second category, and the third category to different dimensional axes of the multidimensional feature space to form a feature vector point; Obtaining an artificial intelligence model; wherein the artificial intelligence model is an autoencoder based on deep learning, which pre-learns the distribution of feature vector points of a known type of camera module in a multidimensional feature space to obtain potential distribution patterns and laws of the feature space; Inputting the feature vector points of the camera module into the artificial intelligence model, and calculating the reconstruction error of the feature vector points in the feature space based on the autoencoder; Determining whether the reconstruction error is within a threshold range; If the reconstruction error is within the threshold range, the closest known type is found as the target classification based on the feature mapping relationship inside the autoencoder and the similarity measurement with the known type; If the reconstruction error exceeds the threshold range, it is judged as a new type or an abnormal type, and further manual intervention mechanism is triggered.

6. The method for identifying camera modules based on artificial intelligence according to claim 1, characterized in that: After the first classification, the second classification and the third classification are fused and analyzed based on the artificial intelligence model to identify the target classification of the camera module, the method includes: Encoding the target classification to obtain a first code; Performing hash operations on the appearance image, the image acquisition method, and the characteristic attributes to obtain corresponding first hash values, second hash values, and third hash values; Reconstructing the first hash value, the second hash value, and the third hash value to generate a hash value matrix; The hash value matrix is ​​screened based on the first code to generate an identification code for identifying images subsequently captured by the camera module.

7. A device for identifying camera modules based on artificial intelligence, characterized in that: include: An acquisition unit, used for acquiring an appearance image of the camera module; Get the image acquisition method of the camera module; Acquire a target image captured by a camera module, and extract characteristic attributes of the target image; An analyzing unit, configured to analyze a first classification of the camera module based on the appearance image; Based on the image acquisition method, analyzing the second classification of the camera module; based on the characteristic attribute, analyzing the third classification of the camera module; An identification unit is used to perform a fusion analysis on the first classification, the second classification and the third classification based on an artificial intelligence model to identify the target classification of the camera module.

8. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

Images