Digital image processing method for multimedia terminal

Abstract

The invention relates to the technical field of image processing, in particular to a digital image processing method for a multimedia terminal, and the method comprises the steps: collecting behavior data and feedback information of a user using an image processing function on multimedia terminal equipment, and carrying out the optimization adjustment of an image processing algorithm through a machine learning model; the method comprises the following specific steps: detecting environment illumination intensity through a light sensor, preprocessing an image, and performing image enhancement, color correction, contrast adjustment and other operations based on a specific algorithm; according to the system, operation behaviors and feedback data of a user are collected in real time, after data cleaning and analysis, an image processing algorithm is optimized by adopting a supervised learning or reinforcement learning model, and image processing parameters are dynamically adjusted according to an actual application scene; according to the invention, the personalized and intelligent level of image processing can be improved, the user experience is remarkably improved, and the method has continuous improvement capability.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a digital image processing method for a multimedia terminal. Background Art

[0002] With the popularization of multimedia terminal devices and the diversification of application scenarios, users' demand for image processing is increasing. Whether on smartphones, tablets, smart TVs, or other terminal devices that support image acquisition and display, image processing technology has become one of the important factors in improving user experience. Users not only want to obtain high-quality images, but also want to make the image processing effect meet their needs through personalized settings and intelligent adjustments.

[0003] Existing image processing technologies mainly rely on fixed algorithms and parameters, lacking the ability to deeply learn and dynamically optimize user feedback and usage habits. In practice, this leads to insufficient generalization of image processing effects and the inability to adapt to different application scenarios and individual user needs. In addition, existing technologies cannot effectively dynamically adjust image processing parameters based on real-time user operations and feedback, resulting in poor system responsiveness and a poor user experience. Summary of the invention

[0004] The invention provides a digital image processing method for a multimedia terminal.

[0005] A digital image processing method for a multimedia terminal comprises the following steps:

[0006] S1, Image Capture and Feature Recognition: Scan and capture the image on the screen through the camera of the multimedia terminal, and apply image processing algorithms to identify important objects and graphic features in the image;

[0007] S2, image segmentation and anti-counterfeiting processing: segment the recognized image into multiple regions, and embed anti-counterfeiting logos and feature data watermarks into the segmented images;

[0008] S3, Adaptive Content Adjustment: Dynamically adjust the parameters of the image processing algorithm to optimize the color performance and contrast of the image based on the captured image content and ambient lighting conditions;

[0009] S4, encoding and compression: Encoding and compressing the anti-counterfeiting processed image to reduce the data transmission volume and ensure that the image quality is not lost during network transmission;

[0010] S5, real-time transmission: through encryption transmission technology, high-quality image remote real-time transmission is achieved, while real-time monitoring is carried out to ensure information security and integrity;

[0011] S6, Intelligent Identification and Matching: Apply intelligent identification algorithms to distinguish and match people and objects in different application scenarios to improve processing efficiency;

[0012] S7, dynamic optimization based on feedback: By collecting user usage data and feedback, machine learning methods are used to dynamically optimize the processing algorithm to continuously improve image processing effects and user experience.

[0013] Optionally, the S1 includes:

[0014] S11, image acquisition: scan and capture the image on the screen in real time through the camera of the multimedia terminal to generate digital image data;

[0015] S12, preprocessing: performing denoising and color correction on the collected image data to generate a corrected image;

[0016] S13, feature extraction: using a convolutional neural network to extract features from the corrected image and output a feature vector F;

[0017] S14, important object recognition: Use the trained support vector machine classification model to classify the extracted feature vectors, determine the important objects in the image, and output the classification results.

[0018] Optionally, S2 includes:

[0019] S21, image segmentation: segment the identified image using an adaptive threshold segmentation algorithm to generate multiple regions, each region representing a different part of the image;

[0020] S22, regional feature extraction: extracting feature data from each segmented region, including color feature extraction and texture feature extraction;

[0021] S23, embedding an anti-counterfeiting mark: embedding an anti-counterfeiting mark in each segmented area according to the extracted regional features;

[0022] S24, feature watermark embedding: embedding feature data watermarks in images to trace the source of the images and prevent computational manipulation.

[0023] Optionally, the S3 includes:

[0024] S31, ambient light intensity detection: detecting the ambient light intensity through a built-in light sensor or camera module of the multimedia terminal, and passing the light intensity as an input parameter to the image processing algorithm;

[0025] S32, image feature analysis: performing feature analysis on the captured image, including color histogram analysis and degree calculation;

[0026] S33, dynamic parameter adjustment: dynamically adjusting the parameters of the image processing algorithm according to the ambient light intensity L and the image feature analysis results, including brightness adjustment and contrast enhancement;

[0027] S34, color performance optimization: adjust the color performance of the image according to the distribution of the color histogram, including color balance and saturation adjustment.

[0028] Optionally, the S4 includes:

[0029] S41, image block processing: dividing the processed image into multiple image blocks;

[0030] S42, discrete cosine transform: perform discrete cosine transform on each image block to convert the pixel value in the spatial domain into the frequency domain representation;

[0031] S43, quantization processing: applying a preset quantization matrix to quantize the discrete cosine transform coefficients;

[0032] S44, entropy coding: Huffman coding is used to entropy code the quantized discrete cosine transform coefficients;

[0033] S45, coded data encapsulation: encapsulating the entropy coded image data for network transmission.

[0034] Optionally, the S5 includes:

[0035] S51, encryption data preparation: encrypting the image data, using the symmetric encryption algorithm AES to encrypt the data;

[0036] S52, data segmentation and transmission: segmenting the encrypted data to adapt to network transmission;

[0037] S53, real-time monitoring and retransmission mechanism: During the data transmission process, the transmission status is monitored in real time and a retransmission mechanism is provided to ensure data integrity and real-time performance;

[0038] S54, data decryption and recovery: after the receiving end completes receiving the image data, the encrypted data is decrypted using the key to restore the original image data;

[0039] S55, image display and feedback: After the receiving end completes the decryption and recovery of the image data, the image is displayed in real time.

[0040] Optionally, the S6 includes:

[0041] S61, scene classification: using a scene classification algorithm to classify the captured image to determine which application scene the image belongs to;

[0042] S62, face and object detection: Based on the scene classification results, dynamically select the target detection model suitable for the current scene, and detect people and objects in the image respectively, including:

[0043] Face detection: Use the YOLO model based on deep learning to detect faces in images and output the bounding box and confidence level of the face;

[0044] Object detection: Use the FasterR-CNN model to detect objects in the image and output the object’s bounding box and category label;

[0045] S63, feature matching and association: based on the detected bounding box and the corresponding feature vector extracted, matching and association are performed;

[0046] S64, dynamic adjustment and optimization: dynamically adjust the parameters of the recognition algorithm according to different application scenarios and user needs.

[0047] Optionally, the S7 includes:

[0048] S71, user data collection: collect the behavior data and feedback information generated by users during use in real time through the interactive interface and background system of the multimedia terminal;

[0049] S72, data preprocessing: cleaning and standardizing the collected user behavior data and feedback data;

[0050] S73, Feature extraction and analysis: Use feature extraction algorithms to extract features from user behavior data and feedback data.

[0051] Optionally, the S7 further includes:

[0052] S74, machine learning model training: using the collected user data and feedback data, using a supervised learning model for training to generate an optimization model for dynamically adjusting the image processing algorithm;

[0053] S75, dynamic adjustment: according to the trained optimization model, the parameters of the image processing algorithm are adjusted dynamically in real time.

[0054] Beneficial effects of the present invention:

[0055] The present invention, through a dynamic optimization method based on user usage data and feedback, can adjust the core parameters of the image processing algorithm (such as brightness, contrast, color saturation, etc.) in real time, so that the image processing effect is more in line with the user's personalized needs. The introduction of machine learning models ensures that the system can adapt to the usage habits of different users, provide a more accurate and intelligent image processing experience, and greatly improve user satisfaction.

[0056] The present invention collects user behavior data and feedback information, and uses supervised learning and reinforcement learning algorithms to automatically optimize image processing, reducing the need for manual adjustment. By regularly updating and retraining the optimization model, it can continuously improve with user needs and environmental changes, realizes automated continuous learning, and ensures the reliability and cutting-edge nature of image processing effects in long-term use.

[0057] The present invention can quickly respond to user operations and feedback through real-time dynamic adjustment of the image processing algorithm, and the adjustment process of the optimization parameters is smooth and efficient, ensuring the rapid response capability of the system in practical applications. At the same time, based on the feedback mechanism of reinforcement learning, the image processing parameters can be accurately optimized according to the actual application scenario and user feedback, ensuring the overall efficiency and effect of image processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 A schematic diagram of a method flow chart of an embodiment of the present invention;

[0060] Figure 2 Schematic diagram of S2 process of an embodiment of the present invention. DETAILED DESCRIPTION

[0061] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0062] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0063] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0064] like Figure 1-Figure 2 As shown, a digital image processing method for a multimedia terminal comprises the following steps:

[0065] S1, Image Capture and Feature Recognition: Scan and capture the image on the screen through the camera of the multimedia terminal, and apply image processing algorithms to identify important objects and graphic features in the image;

[0066] S2, image segmentation and anti-counterfeiting processing: segment the recognized image into multiple regions, and embed anti-counterfeiting logos and feature data watermarks into the segmented images;

[0067] S3, Adaptive Content Adjustment: Dynamically adjust the parameters of the image processing algorithm to optimize the color performance and contrast of the image based on the captured image content and ambient lighting conditions;

[0068] S4, encoding and compression: Encoding and compressing the anti-counterfeiting processed image to reduce the data transmission volume and ensure that the image quality is not lost during network transmission;

[0069] S5, real-time transmission: through encryption transmission technology, high-quality image remote real-time transmission is achieved, while real-time monitoring is carried out to ensure information security and integrity;

[0070] S6, Intelligent Identification and Matching: Apply intelligent identification algorithms to distinguish and match people and objects in different application scenarios to improve processing efficiency;

[0071] S7, dynamic optimization based on feedback: By collecting user usage data and feedback, machine learning methods are used to dynamically optimize the processing algorithm to continuously improve image processing effects and user experience.

[0072] Optionally, the S1 includes:

[0073] S11, image acquisition: Scan and capture the image on the screen in real time through the camera of the multimedia terminal to generate digital image data I, which is a set of pixel values ​​represented in the form of a two-dimensional matrix;

[0074] S12, preprocessing: performing denoising and color correction on the collected image data to generate a corrected image Icolor ;

[0075] A median filter or a Gaussian filter is used to remove image noise to improve image quality. The denoised image is denoted as I denoise , according to the preset color standard, adjust the color saturation and brightness of the image to generate the corrected image I color ;

[0076] S13, feature extraction: Convolutional neural network is used to extract the corrected image I color Perform feature extraction and output feature vector F, where F represents the important graphic features in the image, expressed as:

[0077] F = CNN(I color );

[0078] Among them, CNN represents convolutional neural network, and the final feature vector is obtained after multiple layers of convolution and pooling operations;

[0079] S14, important object identification: Use the trained support vector machine classification model M to classify the extracted feature vector F, determine the important objects O in the image, and output the classification result R, which is expressed as:

[0080] R = M(F);

[0081] Where M is a trained support vector machine classification model used to identify important objects and their categories in the image;

[0082] Among them, the steps of establishing the support vector machine classification model include:

[0083] Training data preparation: Use annotated image dataset {(F i ,O i )}, where F i is the eigenvector, O i is the corresponding category label;

[0084] Model training: Use the support vector machine algorithm to train by optimizing the following objective function:

[0085]

[0086] Among them, w is the weight vector, C is the penalty parameter, ξ i is a slack variable that satisfies the constraint y i (w·F i +b)≥1-ξ i .

[0087] Optionally, S2 includes:

[0088] S21, image segmentation: segment the identified image and use an adaptive threshold segmentation algorithm to generate multiple regions R 1 ,R 2 ,…,R n ,Each region represents a different part of the image;

[0089] Adaptive threshold segmentation: Dynamically adjust the threshold T according to the grayscale distribution of the image so that different thresholds can be applied to segmentation in different areas. The threshold T is calculated as:

[0090]

[0091] Among them, mean bg is the mean of the background pixels, mean fg is the mean value of foreground pixels;

[0092] S22, regional feature extraction: After segmentation, each region R i (where i is 1 to n), including color feature extraction and texture feature extraction, wherein;

[0093] Color feature extraction: Calculate the color histogram H(R) of each region i ), used to identify the color distribution characteristics within the area;

[0094] Texture feature extraction: The gray level co-occurrence matrix is ​​used to calculate the texture feature T(R i ), including contrast, energy, and correlation;

[0095] S23, anti-counterfeiting logo embedding: Based on the extracted regional features, i Embedded anti-counterfeiting mark W i To ensure the anti-counterfeiting effect of the image, the embedding process includes:

[0096] Anti-counterfeiting mark generation: Generate a unique anti-counterfeiting mark W based on the encryption algorithm i , used for identification within the area;

[0097] Anti-counterfeiting logo embedding: Using digital watermark technology, the anti-counterfeiting logo W i Embed each region R i Inconspicuous areas are embedded to ensure that the image quality is not affected after embedding;

[0098] S24, feature watermark embedding: embedding feature data watermark D in a specific area of ​​the image i , used to trace the source of the image and prevent algorithmic modification, watermark embedding uses DCT (discrete cosine transform) and LSB (least significant bit) embedding method. The specific steps are:

[0099] Discrete cosine transform (DCT): Perform DCT transform on the image data of each region to decompose the image data into frequency components;

[0100] Least significant bit embedding: embed the characteristic data watermark into the least significant bit of the DCT transformed image to ensure the invisibility of the watermark;

[0101] Through image segmentation and anti-counterfeiting processing, effective image segmentation and anti-counterfeiting processing are achieved. While ensuring the image quality, the embedded anti-counterfeiting logo and characteristic data watermark improve the authenticity and traceability of the image and enhance the anti-counterfeiting performance of the image.

[0102] Optionally, the S3 includes:

[0103] S31, ambient light intensity detection: detecting the ambient light intensity L through a built-in optical sensor or camera module of the multimedia terminal, and passing the light intensity L as an input parameter to the image processing algorithm;

[0104] S32, image feature analysis: performing feature analysis on the captured image, including color histogram analysis and degree calculation, wherein;

[0105] Color histogram analysis: Calculate the color histogram H(I) of the image to determine the balance of color distribution in the image;

[0106] To calculate this degree: Calculate the contrast C of the image based on the image brightness channel using the following formula:

[0107]

[0108] Among them, I brightness is the brightness component of the image;

[0109] S33, dynamic parameter adjustment: dynamically adjusting the parameters of the image processing algorithm according to the ambient light intensity L and the image feature analysis results, including brightness adjustment and contrast enhancement, wherein;

[0110] Brightness adjustment: When the ambient light intensity is low, increase the image brightness B. The calculation formula is:

[0111] B′=B+k 1 ·(L ref -L);

[0112] Among them, B is the original brightness, B′ is the adjusted brightness, and L ref is the reference light intensity, k 1 is the brightness adjustment coefficient;

[0113] Contrast enhancement: When the image contrast is low, the contrast stretching algorithm is used to adjust the contrast C. The calculation formula is:

[0114] C′=k 2 C;

[0115] Among them, C′ is the adjusted contrast, k 2 is the contrast adjustment factor;

[0116] S34, color performance optimization: adjusting the color performance of the image according to the distribution of the color histogram H(I), including color balance and saturation adjustment, wherein;

[0117] Color balance: When the color distribution of an image is uneven, histogram equalization technology is used to optimize the color distribution to make the color performance of the image more uniform;

[0118] Saturation adjustment: Dynamically adjust the saturation S of the image according to user preferences or scene requirements. Saturation adjustment is expressed as:

[0119] s′=s+k 3 ·(S ref -s);

[0120] Among them, S′ is the adjusted saturation, S is the original saturation, and S ref is the reference saturation, k 3 is the saturation adjustment factor;

[0121] The image processing parameters are adaptively adjusted according to the ambient lighting conditions and image content to optimize the brightness, contrast and color performance, ensuring that the image maintains the best visual effect under different lighting conditions and improving the user's visual experience.

[0122] Optionally, the S4 includes:

[0123] S41, image block processing: dividing the processed image into multiple image blocks I block ,The size of each image block is 8×8 pixels, which is convenient for subsequent encoding and compression operations;

[0124] S42, discrete cosine transform: for each image block I block Perform discrete cosine transform to convert the pixel value in the spatial domain into the frequency domain representation, which is expressed as:

[0125]

[0126] Among them, F(u,v) is the frequency domain coefficient, I block (x, y) is the pixel value of the image block;

[0127] S43, quantization processing: applying a preset quantization matrix Q to quantize the discrete cosine transform coefficients F(u,v). The quantized coefficients F q (u,v) is expressed as:

[0128]

[0129] Among them, Q(u,v) is the element of the quantization matrix, which controls the degree of quantization. This step reduces the accuracy of the high-frequency part, thereby achieving image compression;

[0130] S44, entropy coding: Huffman coding is used to encode the quantized discrete cosine transform coefficients F q (u,v) is entropy coded, and entropy coding further compresses image data by assigning variable-length codes to symbols of different frequencies. The specific steps are as follows:

[0131] (1) Symbol statistics: Count the frequency of occurrence of each symbol in the quantization coefficient;

[0132] (2) Huffman tree generation: Generate a Huffman tree based on the symbol frequency and assign a corresponding variable-length code to each symbol;

[0133] (3) Encoded data generation: Convert the quantized coefficients into corresponding Huffman codes to generate the final encoded data C d ;

[0134] S45, coded data packaging: encapsulate the entropy coded image data C d Encapsulation is performed to facilitate network transmission. The encapsulation process includes adding data header information and adjusting the data format to ensure that the data can be correctly parsed during transmission;

[0135] The image is efficiently compressed using multiple steps such as blocking, discrete cosine transform, quantization and entropy coding, ensuring that the image quality is retained to the maximum extent while reducing the amount of data transmission, so that the visual effect of the image is not lost during network transmission.

[0136] Optionally, the S5 includes:

[0137] S51, encryption data preparation: Encrypt the image data, use the symmetric encryption algorithm AES (Advanced Encryption Standard) to encrypt the data, the encryption steps are as follows:

[0138] Key generation: Generate a 256-bit key K for encrypting and decrypting image data;

[0139] Image data encryption: The compressed image data C d As input, the image data is encrypted using the AES algorithm using the key K to obtain the encrypted data E d , the formula is:

[0140] E d =AES K (Cd );

[0141] Among them, AES K Indicates AES encryption using key K;

[0142] S52, data segmentation and transmission: The encrypted data E d Slice to adapt to network transmission, including the following steps

[0143] Data sharding: Encrypted data E d According to the preset transmission packet size, it is fragmented to generate multiple data fragments E d1 ,E d2 ,…,E dn ,The size of each fragment depends on the requirements of the network transmission protocol (e.g., the maximum transmission unit MTU of TCP / IP packets);

[0144] Transport protocol selection: Select an appropriate transport protocol according to the transmission requirements, such as UDP (User Datagram Protocol) or TCP (Transmission Control Protocol) to ensure real-time transmission and stability of data. UDP is suitable for scenarios with high real-time requirements;

[0145] S53, real-time monitoring and retransmission mechanism: During the data transmission process, real-time monitoring of the transmission status and provision of a retransmission mechanism ensures data integrity and real-time performance, including:

[0146] Transmission status monitoring: real-time monitoring of the sending status of data fragments, recording the sending time, sending result (success or failure) and network delay of each data fragment;

[0147] Data integrity check: At the receiving end, the received data fragments are checked for integrity using the MD5 or SHA-256 hash algorithm to generate a hash value H recv , and the hash value H of the sender sent Compare to ensure that the data has not been altered. The formula is:

[0148] H recv =H sent ;

[0149] Data retransmission mechanism: If a transmission error or data loss is found, the retransmission mechanism is started and the lost data fragment is immediately re-requested. di ,Ensure transmission integrity;

[0150] S54, data decryption and recovery: After the receiving end completes the reception of the image data, the encrypted data is decrypted using the key to restore the original image data C d , expressed as:

[0151]

[0152] in, Indicates AES decryption using key K;

[0153] S55, image display and feedback: after the receiving end completes the decryption and recovery of the image data, the image is displayed in real time;

[0154] The use of AES encryption, data segmentation, transmission monitoring and retransmission mechanisms ensures the security, integrity and real-time performance of image data during remote real-time transmission, and realizes encrypted transmission and recovery of high-quality images.

[0155] Optionally, the S6 includes:

[0156] S61, scene classification: Use scene classification algorithm to classify the captured image to determine which application scene the image belongs to. The scene classification algorithm is trained with convolutional neural network (CNN) to identify different application scenes such as indoor, outdoor, transportation, office, etc., and output the classification result S c , expressed as:

[0157] S c =CNN(I);

[0158] Among them, I is the input image data, S c The result of scene classification;

[0159] S62, face and object detection: based on the scene classification results S c , dynamically select the target detection model suitable for the current scene, and detect people and objects in the image respectively, including:

[0160] Face detection: Use the YOLO (You Only Look Once) model based on deep learning to detect faces in the image and output the bounding box B of the face. f and confidence C f , the formula is:

[0161] B f ,C f =YOLO face (I);

[0162] Among them, B f is the bounding box of the face, C f is the confidence level;

[0163] Object detection: Use the FasterR-CNN model to detect objects in the image and output the bounding box B of the object. o And the category label O, expressed as:

[0164] B o ,O=Faster R-CNN(I);

[0165] Among them, B o is the bounding box of the object, and O is the category label of the object;

[0166] S63, Feature matching and association: Based on the detected bounding box B f and B o Extract the corresponding feature vectors, match and associate them, and ensure that the relationship between people and objects can be identified and associated. The specific steps are:

[0167] Feature extraction: A convolutional neural network is used to extract features from the image area within the bounding box to obtain the feature vector F of people and objects. f and F o , the formula is:

[0168]

[0169] Feature matching: Calculate the feature matching between people and objects based on Euclidean distance or cosine similarity. The matching result M fo Calculated as:

[0170]

[0171] Among them, M fo It is the similarity matching value between people and objects. The higher the matching degree, the stronger the association.

[0172] S64, Dynamic Adjustment and Optimization: Dynamically adjust the parameters of the recognition algorithm according to different application scenarios and user needs to ensure efficient and accurate recognition and matching in various scenarios, including:

[0173] Parameter optimization: Optimize detection accuracy and speed by adjusting the confidence threshold and non-maximum suppression (NMS) parameters of the target detection model;

[0174] Scene adaptive adjustment: According to the scene classification result S c ,Dynamically adjust the model selection and parameter configuration of object recognition to meet the needs of different application scenarios;

[0175] Based on intelligent recognition algorithms, it effectively distinguishes and matches people and objects in images, ensuring efficient object detection and association in different application scenarios, significantly improving processing efficiency and accuracy.

[0176] Optionally, the S7 includes:

[0177] S71, user data collection: collect the behavior data and feedback information generated by users during use in real time through the interactive interface and background system of the multimedia terminal, including:

[0178] User behavior data record: record the user's operation behavior on image processing, including image enhancement, color adjustment, contrast adjustment and other operation parameters, data set D u By operation record O r And the corresponding parameter P;

[0179] D u ={(O r ,P)};

[0180] User feedback data collection: Through the user evaluation system, collect user feedback on image processing effects, such as image clarity, color performance, operation fluency, etc., to generate feedback data set F u ;

[0181] S72, data preprocessing: clean and standardize the collected user behavior data and feedback data to ensure the accuracy and consistency of the data. The specific steps include:

[0182] Data denoising: Remove invalid or noisy data to ensure that the data used for analysis is of high quality;

[0183] Data standardization: standardize parameters of different units or ranges into a unified numerical range to facilitate subsequent algorithm processing;

[0184] S73, Feature extraction and analysis: Use feature extraction algorithms to extract features from user behavior data and feedback data, including:

[0185] Behavior feature extraction: cluster analysis of user operation behaviors, identify common operation modes, and generate operation mode features F o ;

[0186] F o =Cluster(D u );

[0187] Feedback analysis: Perform sentiment analysis and score summary on user feedback data to generate a user satisfaction score S u ,The rating range is from 1 to 10, indicating the satisfaction with the ,image processing effect.

[0188] Optionally, the S7 further includes:

[0189] S74, machine learning model training: using the collected user data and feedback data, using the supervised learning model for training, and generating an optimization model M for dynamically adjusting the image processing algorithm opt, the specific steps include:

[0190] Supervised learning model training: based on user behavior data D u and feedback score S u ,The parameters of the image processing algorithm are optimized by using regression models (such as linear regression and decision tree regression), and the optimization objective function is:

[0191]

[0192] in, Represents the user's feedback rating. Indicates operation record, P i Indicates operation parameters;

[0193] S75, dynamic adjustment: according to the trained optimization model M opt , the parameters of the image processing algorithm are adjusted dynamically in real time. The specific steps include:

[0194] Real-time parameter optimization: Based on the optimal parameter combination output by the model, the core algorithm parameters of image processing (such as brightness, contrast, and color saturation) are automatically adjusted to ensure that the image processing effect can adapt to user needs;

[0195] Continuous learning and updating: The system regularly retrains and updates the model based on new user behavior data and feedback to ensure that the optimization strategy can be continuously adjusted as user habits and needs change.

[0196] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0197] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A digital image processing method for a multimedia terminal, characterized in that: The following steps are involved: S1, Image Capture and Feature Recognition: Scan and capture the image on the screen through the camera of the multimedia terminal, and apply image processing algorithms to identify important objects and graphic features in the image; S2, image segmentation and anti-counterfeiting processing: segment the recognized image into multiple regions, and embed anti-counterfeiting logos and feature data watermarks into the segmented images; S3, Adaptive Content Adjustment: Dynamically adjust the parameters of the image processing algorithm based on the captured image content and ambient lighting conditions; S4, encoding and compression: encoding and compressing the anti-counterfeiting processed image to reduce the amount of data transmission; S5, real-time transmission: through encrypted transmission technology, remote real-time transmission of images is realized, and real-time monitoring is carried out at the same time; S6, Intelligent Identification and Matching: Apply intelligent identification algorithms to distinguish and match people and objects in different application scenarios; S7, dynamic optimization based on feedback: By collecting user usage data and feedback, machine learning methods are used to dynamically optimize the processing algorithm.

2. A digital image processing method for a multimedia terminal according to claim 1, characterized in that: The S1 includes: S11, image acquisition: scan and capture the image on the screen in real time through the camera of the multimedia terminal to generate digital image data; S12, preprocessing: performing denoising and color correction on the collected image data to generate a corrected image; S13, feature extraction: using a convolutional neural network to extract features from the corrected image and output a feature vector F; S14, important object recognition: Use the trained support vector machine classification model to classify the extracted feature vectors, determine the important objects in the image, and output the classification results.

3. A digital image processing method for a multimedia terminal according to claim 2, characterized in that: The S2 includes: S21, image segmentation: segment the identified image using an adaptive threshold segmentation algorithm to generate multiple regions, each region representing a different part of the image; S22, regional feature extraction: extracting feature data from each segmented region, including color feature extraction and texture feature extraction; S23, embedding an anti-counterfeiting mark: embedding an anti-counterfeiting mark in each segmented area according to the extracted regional features; S24, feature watermark embedding: embedding feature data watermarks in images to trace the source of the images and prevent computational manipulation.

4. A digital image processing method for a multimedia terminal according to claim 3, characterized in that: The S3 includes: S31, ambient light intensity detection: detecting the ambient light intensity through a built-in light sensor or camera module of the multimedia terminal, and passing the light intensity as an input parameter to the image processing algorithm; S32, image feature analysis: performing feature analysis on the captured image, including color histogram analysis and degree calculation; S33, dynamic parameter adjustment: dynamically adjusting the parameters of the image processing algorithm according to the ambient light intensity and the image feature analysis results, including brightness adjustment and contrast enhancement; S34, color performance optimization: adjust the color performance of the image according to the distribution of the color histogram, including color balance and saturation adjustment.

5. A digital image processing method for a multimedia terminal according to claim 4, characterized in that: The S4 includes: S41, image block processing: dividing the processed image into multiple image blocks; S42, discrete cosine transform: perform discrete cosine transform on each image block to convert the pixel value in the spatial domain into the frequency domain representation; S43, quantization processing: applying a preset quantization matrix to quantize the discrete cosine transform coefficients; S44, entropy coding: Huffman coding is used to entropy code the quantized discrete cosine transform coefficients; S45, coded data encapsulation: encapsulating the entropy coded image data for network transmission.

6. A digital image processing method for a multimedia terminal according to claim 5, characterized in that: The S5 includes: S51, encryption data preparation: encrypting the image data, using the symmetric encryption algorithm AES to encrypt the data; S52, data segmentation and transmission: segmenting the encrypted data to adapt to network transmission; S53, real-time monitoring and retransmission mechanism: During the data transmission process, the transmission status is monitored in real time and a retransmission mechanism is provided to ensure data integrity and real-time performance; S54, data decryption and recovery: after the receiving end completes receiving the image data, the encrypted data is decrypted using the key to restore the original image data; S55, image display and feedback: After the receiving end completes the decryption and recovery of the image data, the image is displayed in real time.

7. A digital image processing method for a multimedia terminal according to claim 6, characterized in that: The S6 includes: S61, scene classification: using a scene classification algorithm to classify the captured image to determine which application scene the image belongs to; S62, face and object detection: Based on the scene classification results, dynamically select the target detection model suitable for the current scene, and detect people and objects in the image respectively, including: Face detection: Use the YOLO model based on deep learning to detect faces in images and output the bounding box and confidence level of the face; Object detection: Use the FasterR-CNN model to detect objects in the image and output the object’s bounding box and category label; S63, feature matching and association: based on the detected bounding box and the corresponding feature vector extracted, matching and association are performed; S64, dynamic adjustment and optimization: dynamically adjust the parameters of the recognition algorithm according to different application scenarios and user needs.

8. A digital image processing method for a multimedia terminal according to claim 7, characterized in that: The S7 includes: S71, user data collection: collect the behavior data and feedback information generated by users during use in real time through the interactive interface and background system of the multimedia terminal; S72, data preprocessing: cleaning and standardizing the collected user behavior data and feedback data; S73, Feature extraction and analysis: Use feature extraction algorithms to extract features from user behavior data and feedback data.

9. A digital image processing method for a multimedia terminal according to claim 8, characterized in that: The S7 further comprises: S74, machine learning model training: using the collected user data and feedback data, using a supervised learning model for training to generate an optimization model for dynamically adjusting the image processing algorithm; S75, dynamic adjustment: according to the trained optimization model, the parameters of the image processing algorithm are adjusted dynamically in real time.

Images