A video encoding and decoding method and device based on intelligent digital retina

By using depth models to perform feature extraction and entropy encoding on video images in intelligent digital retinal technology, the feature enhancement stream and basic stream are generated, which solves the problem of high-cost computing consumption in the existing technology, and realizes high-resolution playback and optimization Video compression performance.

CN114630129BActive Publication Date: 2024-12-03ZHEJIANG SMART VIDEO SECURITY INNOVATION CENT CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210116067.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-02-07
Publication Date
2024-12-03
Estimated Expiration
2042-02-07

AI Technical Summary

Technical Problem

The existing video processing methods based on intelligent digital retina have a large amount of additional computing consumption, resulting in large storage overhead, making it difficult to obtain as much video data processing capability as possible with as little computing consumption as possible.

Method used

The video image is characterized by extracting and entropy encoding with the first preset depth model, generating feature enhancement streams and basic streams, and obtaining the encoded video image through multiplexing, optimizing the compression and playback performance of the video.

Benefits of technology

High-resolution video playback under the same transmission and storage resources are realized, image feature extraction performance and picture quality are improved, computing overhead is reduced, and video compression and playback performance is optimized.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114630129B_ABST
    Figure CN114630129B_ABST
Patent Text Reader

Abstract

The present invention discloses a video encoding and decoding method and device based on intelligent digital retina, wherein the video encoding method includes: entropy encoding a first prediction residual to obtain a corresponding first feature enhancement stream; and multiplexing the first base stream and the first feature enhancement stream to obtain an encoded video image. The video encoding method provided in the embodiment of the present application processes the video image by introducing a first preset depth model. Since the first preset depth model can extract features of the image, accurately determine the feature area of ​​the video image, and perform entropy encoding based on the first prediction residual to obtain a first feature enhancement stream, and obtain the encoded video image based on the first feature enhancement stream and the first base stream; in this way, the encoded video image has better picture quality; in addition, the first preset depth model can also optimize the compression performance and playback performance of the video.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of video coding technology, and in particular to a video coding and decoding method and device based on intelligent digital retina. Background Art

[0002] Since the concept of digital retina was proposed, it has attracted great attention in the fields of video encoding and decoding, video surveillance, etc. In the traditional field of image processing, video compression and video analysis belong to two different fields. Inspired by the biological functions of the human retina, digital retina technology is the first to propose an intelligent image sensor that integrates video compression and video analysis. Specifically, the characteristic of digital retina is that it can obtain video compression data and video feature data at the same time, and transmit them to the cloud through data streams, which is convenient for later playback and retrieval. In order to obtain the feature stream of the image, digital retina technology introduces the concept of model stream, that is, the image acquisition front end can apply different feature extraction models according to needs, and these models can be sent to the image acquisition front end through cloud storage and reverse transmission. In terms of video compression, the basic concept is to compress the spatiotemporal redundant information of the video by calculation. The basic paradigm of video compression has not changed much in the past decades. Block-based video compression codec technology has developed very maturely, with the characteristics of moderate computational complexity, high compression rate, and high reconstruction quality. Therefore, it has been widely used in the past decades. The current mainstream codec technologies include H.264 / H.265 / H.266 and MPEG2 / MPEG4, which are mainly based on block-based video codec technology. Since the early days of video coding, the paradigm of coding theory has not changed. The technologies used by the new generation of coding standards are all to improve the compression ratio by the method of "computation for space". For example, the evolution from H.264 to H.265 has increased the compression rate by 50%, but it also brings greater computing requirements. This is due to the use of more flexible coding units and more flexible reference frames, which enable motion compensation-based compression methods to tap more compression potential.

[0003] Since the digital retina framework integrates the two aspects of video-related feature recognition and data compression, a new paradigm has been created. This paradigm excludes a certain technology measured by a single parameter, but instead uses a comprehensive evaluation method oriented to complex goals. This is also a valuable inspiration obtained from the biological structure of the retina. The retina is not simply transmitting or compressing image data, but an intelligent front-end image acquisition and analysis device that serves various complex tasks of the brain. The digital retina has created a new thinking paradigm, which is a method of developing various technical means with a human-centered approach, rather than developing new technologies based on images centered on a single indicator of image processing.

[0004] In some prior arts, although the professional terminology of digital retina is not used, similar concepts are actually followed. For example, in CN111083477A, ​​the inventor proposed a method of using an attention model to adaptively adjust the quantization parameters used in HEVC encoding. For another example, CN107396124B proposes a method for further compressing the prediction residual based on a neural network. In these methods, the introduction of deep models enables the computer to recognize features in the video, and further block-based video encoders can use the analysis results for compression optimization. For example, the method proposed by CN111083477A allows the encoder to use a more optimized quantizer for content that the user is highly sensitive to. However, since these methods are limited by the existing framework of the encoder, the performance gain obtained is at the expense of the huge computational consumption of the deep model.

[0005] How to obtain as much video data processing capability as possible with as little additional computing consumption as possible is a technical problem to be solved. Summary of the invention

[0006] Based on this, it is necessary to provide a video processing method, device, electronic device and storage medium based on an intelligent digital retina to address the problem that the existing video processing method based on the intelligent digital retina consumes a lot of storage overhead.

[0007] In a first aspect, an embodiment of the present application provides a video encoding method based on an intelligent digital retina, the method comprising:

[0008] Acquire an original video image based on the collected video data;

[0009] Inputting the original video image into a first preset depth model for image processing to obtain a corresponding feature sequence for image retrieval, a first feature region image of a first feature region, position information of the first feature region, and a first predicted image with a first preset resolution, wherein the feature sequence includes a feature sequence of a focus feature region;

[0010] Calculating a residual value between the first feature region image and the first predicted image with a first preset resolution to obtain a first prediction residual;

[0011] The first prediction residual is entropy encoded to obtain a corresponding first feature enhancement stream; and the first basic stream and the first feature enhancement stream are multiplexed to obtain an encoded video image.

[0012] In one implementation, the multiplexing of the first base stream and the first feature enhancement stream to obtain an encoded video image includes:

[0013] Downsampling the original video image to obtain a processed video image with a second preset resolution;

[0014] Performing video encoding on the processed video image to obtain the first elementary stream;

[0015] The first basic stream and the first feature enhancement stream are multiplexed to obtain the encoded video image.

[0016] In one embodiment, the method further comprises:

[0017] Reading the position information of the feature area and the feature area image from a first frame buffer;

[0018] The position information and the characteristic region image are synchronously encoded.

[0019] In a second aspect, an embodiment of the present application provides a video decoding method based on an intelligent digital retina, the method comprising:

[0020] In response to a playback instruction from a user to play back a target video, the second frame buffer reads a second feature region of the target video image and generates a corresponding second feature region image;

[0021] Inputting the second feature region image into a second preset depth model for image processing to obtain a corresponding third feature region image with a third preset resolution;

[0022] Performing entropy decoding on a second feature enhancement stream corresponding to the second feature area to obtain a corresponding second prediction residual; and performing image synthesis based on the second prediction residual and the third feature area image to generate a first decoded target video image with the third preset resolution.

[0023] In one embodiment, the method further comprises:

[0024] Optimizing the second preset depth model to obtain an optimized depth model, wherein the optimized depth model includes a third preset depth model that performs image optimization processing on a non-feature area of ​​a target video image;

[0025] Based on the optimized depth model, the target video image is decoded to generate a second decoded target video image with the third preset resolution.

[0026] In one implementation, the decoding the target video image based on the optimized depth model to generate a second decoded target video image having the third preset resolution includes:

[0027] The second frame buffer generates a non-feature area image based on the second basic stream;

[0028] Inputting the non-feature area image into a third preset depth model for image processing to obtain a first target video image of the non-feature area having the third preset resolution;

[0029] Acquire a second target video image having a characteristic area of ​​the target video image and the third preset resolution;

[0030] The first target video image and the second target video image are subjected to image synthesis processing to generate a second decoded target video image having the third preset resolution.

[0031] In a third aspect, an embodiment of the present application provides a video encoding device based on an intelligent digital retina, the device comprising:

[0032] An acquisition module, used to acquire an original video image based on the collected video data;

[0033] A first image processing module, used for inputting the original video image acquired by the acquisition module into a first preset depth model for image processing, to obtain a corresponding feature sequence for image retrieval, a first feature region image of a first feature region, position information of the first feature region, and a first predicted image with a first preset resolution, wherein the feature sequence includes a feature sequence of a focus feature region;

[0034] a calculation module, configured to calculate a residual value between the first feature region image obtained by the image processing module and the first prediction image with a first preset resolution to obtain a first prediction residual;

[0035] The encoding module is used to perform entropy encoding on the first prediction residual obtained by the calculation module to obtain a corresponding first feature enhancement stream; and multiplex the first basic stream and the first feature enhancement stream to obtain an encoded video image.

[0036] In a fourth aspect, an embodiment of the present application provides a video decoding device based on an intelligent digital retina, the device comprising:

[0037] A reading and generating module, configured to respond to a playback instruction of a user to play back a target video, and to cause a second frame buffer to read a second characteristic region of a target video image and generate a corresponding second characteristic region image;

[0038] A second image processing module, used for inputting the second feature area image obtained by the reading and generating module into a second preset depth model for image processing to obtain a corresponding third feature area image with a third preset resolution;

[0039] A decoding module is used to entropy decode a second feature enhancement stream corresponding to the second feature area to obtain a corresponding second prediction residual; and perform image synthesis based on the second prediction residual and the third feature area image to generate a first decoded target video image with the third preset resolution.

[0040] In a fifth aspect, an embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method steps described above.

[0041] In a sixth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the program is executed by a processor to implement the method steps described above.

[0042] The technical solution provided by the embodiments of the present application may have the following beneficial effects:

[0043] In an embodiment of the present application, an original video image obtained based on the collected video data is obtained; the original video image is input into a first preset depth model for image processing to obtain a corresponding feature sequence for image retrieval, a first feature region image of a first feature region, position information of the first feature region, and a first predicted image with a first preset resolution, wherein the feature sequence includes a feature sequence of a focus feature region; the first feature region image and the first predicted image with a first preset resolution are subjected to residual value calculation to obtain a first prediction residual; the first prediction residual is entropy encoded to obtain a corresponding first feature enhancement stream; and the first basic stream and the first feature enhancement stream are multiplexed to obtain an encoded video image. The video encoding method based on intelligent digital retina provided in an embodiment of the present application processes the video image by introducing a first preset depth model, and since the first preset depth model can extract features of the image features, accurately determine the feature region of the video image, and perform entropy encoding based on the first prediction residual to obtain a first feature enhancement stream, and obtain an encoded video image based on the first feature enhancement stream and the first basic stream; in this way, the encoded video image has better picture quality; in addition, the first preset depth model can also optimize the compression performance and playback performance of the video. It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0045] Figure 1 It is a flowchart of a video encoding method based on intelligent digital retina provided in an embodiment of the present application;

[0046] Figure 2 It is a traditional video compression flow chart based on motion prediction and compensation;

[0047] Figure 3 This is a schematic diagram of a process flow of encoding processing performed by an encoding end in a specific application scenario provided by an embodiment of the present application;

[0048] Figure 4 This is another schematic diagram of a process for encoding processing performed by an encoding end in a specific application scenario provided by an embodiment of the present application;

[0049] Figure 5 This is a flowchart of decoding processing performed by a decoding end in a specific application scenario provided by an embodiment of the present application;

[0050] Figure 6 This is another flowchart of decoding processing performed by a decoding end in a specific application scenario provided by an embodiment of the present application;

[0051] Figure 7 It is a structural schematic diagram of a video encoding device based on an intelligent digital retina provided in an embodiment of the present application;

[0052] Figure 8 A schematic diagram of a connection structure of an electronic device according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0053] The following description and the drawings sufficiently illustrate specific embodiments of the invention to enable those skilled in the art to practice them.

[0054] It should be clear that the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] The optional embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0056] like Figure 1 As shown, it is a flow chart of a video encoding method based on intelligent digital retina provided in an embodiment of the present application; Figure 1 As shown, the embodiment of the present application provides a video encoding method based on intelligent digital retina, which specifically includes the following method steps:

[0057] S102: Acquire an original video image based on the collected video data.

[0058] S104: Input the original video image into the first preset depth model for image processing to obtain a corresponding feature sequence for image retrieval, a first feature area image of the first feature area, location information of the first feature area and a first predicted image with a first preset resolution, wherein the feature sequence includes a feature sequence of a focus feature area.

[0059] like Figure 2 As shown in FIG. 1 , it is a conventional video compression flow chart based on motion prediction and compensation. Where e is the residual value after motion compensation. The residual value is transmitted or stored after passing through a transformer, a quantizer, and an entropy encoder. A common quantizer may be a DCT quantizer. Generally speaking, Figure 2 Video compression is based on a fixed image resolution. In actual devices, the resolution of the image acquisition device is greater than the resolution of the image used for compression, because the data transmission and storage costs are higher. Therefore, the high-resolution image acquired by the image is usually downsampled before video compression. However, from the perspective of playback, the higher the resolution, the better.

[0060] In this step, the first preset depth model has better computing power and can achieve high-resolution playback under the same transmission and storage resources; in addition, the first preset depth model also has better image feature extraction performance.

[0061] The video encoding method provided in the embodiment of the present application is based on intelligent digital retina technology. The principle of the intelligent digital retina technology is as follows:

[0062] The front-end device has both video compression and a deep model for video feature extraction. The back-end can deploy different models to the front-end through transmission methods. Therefore, it can be understood that the front-end device has the ability to adaptively obtain any deep model. Therefore, as long as a model with special feature extraction capabilities is trained offline, it can be deployed to the front-end device through the model stream adopted by the intelligent digital retina technology. In the cloud, the main purpose of the feature stream is to perform image retrieval. After the user obtains the retrieval results, a common linkage requirement is to play back the image or video. At this time, the playback resolution must be as high as possible; that is, the distribution of the feature stream in the video is related to the distribution of the user's expectations for image quality.

[0063] In an embodiment of the present application, the first preset depth model is used to extract features of the current image and output a feature region map. The region map can be a rectangular region. The first preset depth model also outputs a high-resolution predicted image of the region. Further, a high-resolution image of the corresponding region is obtained, and a residual value with the predicted image is calculated. The residual value is encoded by entropy coding.

[0064] In this step, the first preset resolution is a high resolution.

[0065] S106: Calculating a residual value between the first feature region image and the first prediction image with a first preset resolution to obtain a first prediction residual.

[0066] In this step, the first preset resolution is a high resolution.

[0067] S108: performing entropy coding on the first prediction residual to obtain a corresponding first feature enhancement stream; and multiplexing the first basic stream and the first feature enhancement stream to obtain an encoded video image.

[0068] In a possible implementation, multiplexing the first basic stream and the first feature enhancement stream to obtain an encoded video image includes the following steps:

[0069] Downsampling the original video image to obtain a processed video image with a second preset resolution;

[0070] Performing video encoding on the processed video image to obtain a first basic stream;

[0071] The first basic stream and the first feature enhancement stream are multiplexed to obtain an encoded video image.

[0072] In this step, the second preset resolution is a low resolution.

[0073] In a possible implementation, the video encoding method provided in the embodiment of the present application further includes the following steps:

[0074] Reading the position information and the image of the feature region from the first frame buffer;

[0075] The position information and the feature area image are encoded synchronously.

[0076] like Figure 3 The figure is a flow chart of encoding processing performed by the encoding end in a specific application scenario provided by an embodiment of the present application.

[0077] like Figure 3 As shown in the figure, the processing flow of the encoding end is demonstrated: the original image is a high-resolution image, and after sampling, a normal resolution image for encoding is obtained. Further, the image is input into a first preset depth model, and the first preset depth model outputs a feature sequence for image retrieval. In the video encoding method provided in the embodiment of the present application, the first preset depth model outputs a high-resolution image at the same time. It is feasible to set up a separate high-resolution prediction model. By Figure 3In the manner shown, the existing feature extraction model can be reused to obtain a high-resolution predicted image. The feature sequence includes the region of interest. Assuming that the high-resolution predicted image is an M*M image, the image with feature regions is N*N (M>N). Further, the encoder is based on Figure 3 The output of is encoded. Through the above encoding process, the computational overhead can be effectively reduced.

[0078] like Figure 4 As shown, the encoder reads out the feature area from the first frame buffer. At this time, the first preset depth model can also output the location information of the feature area, and the location information of the feature area is also encoded synchronously. Further, the predicted image output by the depth model and the feature area image are calculated to obtain the prediction residual. After the prediction residual is entropy encoded, it is multiplexed with the data stream output by the low-resolution block-based encoder and used for transmission or storage.

[0079] In an embodiment of the present application, an original video image obtained based on the collected video data is obtained; the original video image is input into a first preset depth model for image processing to obtain a corresponding feature sequence for image retrieval, a first feature region image of a first feature region, position information of the first feature region, and a first predicted image with a first preset resolution, wherein the feature sequence includes a feature sequence of a focus feature region; the first feature region image and the first predicted image with a first preset resolution are subjected to residual value calculation to obtain a first prediction residual; the first prediction residual is entropy encoded to obtain a corresponding first feature enhancement stream; and the first basic stream and the first feature enhancement stream are multiplexed to obtain an encoded video image. The video encoding method based on intelligent digital retina provided in an embodiment of the present application processes the video image by introducing a first preset depth model, and since the first preset depth model can extract features of the image features, accurately determine the feature region of the video image, and perform entropy encoding based on the first prediction residual to obtain a first feature enhancement stream, and obtain an encoded video image based on the first feature enhancement stream and the first basic stream; in this way, the encoded video image has better picture quality; in addition, the first preset depth model can also optimize the compression performance and playback performance of the video.

[0080] The embodiment of the present application also provides a video decoding method based on intelligent digital retina, which specifically includes the following method steps:

[0081] S202: In response to a playback instruction from a user to play back a target video, a second frame buffer reads a second feature region of the target video image and generates a corresponding second feature region image.

[0082] S204: Inputting the second feature region image into a second preset depth model for image processing to obtain a corresponding third feature region image having a third preset resolution.

[0083] In this step, the first preset depth model has better computing power and can achieve high-resolution playback under the same transmission and storage resources; in addition, the first preset depth model also has better image feature extraction performance. In addition, the third preset resolution is high resolution.

[0084] S206: performing entropy decoding on a second feature enhancement stream corresponding to the second feature area to obtain a corresponding second prediction residual; and performing image synthesis based on the second prediction residual and the third feature area image to generate a first decoded target video image with a third preset resolution.

[0085] In a possible implementation, the video decoding method provided in the embodiment of the present application further includes the following steps:

[0086] Optimizing the second preset depth model to obtain an optimized depth model, wherein the optimized depth model includes a third preset depth model that performs image optimization processing on a non-feature area of ​​a target video image;

[0087] Based on the optimized depth model, the target video image is decoded to generate a second decoded target video image with a third preset resolution.

[0088] In a possible implementation, decoding the target video image based on the optimized depth model to generate a second decoded target video image with a third preset resolution includes the following steps:

[0089] The second frame buffer generates a non-feature area image based on the second basic stream;

[0090] Inputting the non-feature area image into a third preset depth model for image processing to obtain a first target video image of the non-feature area having a third preset resolution;

[0091] Acquire a second target video image having a feature area of ​​the target video image and a third preset resolution;

[0092] The first target video image and the second target video image are subjected to image synthesis processing to generate a second decoded target video image having a third preset resolution.

[0093] In a specific application scenario, after the user searches based on the feature stream, part of the search results will be played back. At this time, the decoder first decodes the basic stream. When the user chooses to enhance the display of the feature area, the decoder first inputs the stored image into the above model and obtains high-resolution prediction data, and further decodes the residual value of the high-resolution data, and finally displays the high resolution in the target area.

[0094] On the decoding end, users can choose to play back normal videos for browsing, and if necessary, they can also play back the feature area in high resolution. When users need to play back in high resolution, the decoder's workflow is as follows: Figure 5 shown.

[0095] The decoder can reconstruct a high-resolution image with high accuracy in the feature area. However, only a low-resolution image can be reconstructed in the non-feature area. In one embodiment, the decoder can use another depth model to reconstruct a high-resolution image in the non-feature area. The encoder of this method is as follows Figure 4 As shown, the decoding end is Figure 6 As shown. When the decoder reads out the low-resolution buffer, it not only reads out the image of the feature area, but also reads out the image of the non-feature area. The image of the non-feature area is input into the depth model 2 (i.e., the third preset depth model), and then a high-resolution image of the non-feature area is generated. Finally, the high-resolution image of the non-feature area and the high-resolution image of the feature area are synthesized into a complete high-resolution image. Figure 6 The architecture has a great advantage in the application scenarios of digital retina. This is because non-feature areas are often scene images, and the images in these areas are relatively static. Therefore, there is no need to send any relevant data on the encoding side, but rely on a deep model for resolution improvement. Due to the relative static nature of the image, an offline trained model can achieve better performance. On the contrary, the feature area is composed of relatively dynamic targets. Therefore, a prediction model may not be able to obtain better prediction accuracy under all circumstances due to the limitation of training data. Therefore, encoding the residual data and reconstructing it on the decoding side can ensure the reconstruction quality of the feature area. Therefore, using Figure 6 The architecture can achieve overall high-quality image reconstruction at a reasonable computational complexity and transmission cost. In essence, this advantage is brought about by utilizing the feature extraction model to simultaneously complete the two functions of feature extraction and encoding method judgment.

[0096] The video encoding and decoding method based on intelligent digital retina provided in the embodiment of the present application adopts an intelligent digital retina-based architecture, and the model of the video acquisition front end is configurable and can be deployed from the cloud to the video acquisition front end. Therefore, in one embodiment, if Figure 3 as well as Figure 4The encoder side, and Figure 5 and Figure 6 The decoder side can be configured with multiple depth models, and different preset depth models have different feature extraction capabilities and image prediction capabilities. The outputs of different preset depth models on the encoding side are encoded into the video stream. Since each preset depth model can be independently reconstructed based on the decoded output, high-resolution image reconstruction for different tasks can be achieved by calling different models in the cloud. For example, in the same video stream, the user played back the pedestrians in the monitoring area with high resolution at the first time, and the decoder called the preset depth model A; the user played back the cars in the monitoring area with high resolution at the second time, and the decoder called the preset depth model B. In this way, new preset depth models can be continuously trained, so that the system can gradually obtain enhanced performance under the same hardware architecture.

[0097] The following is an embodiment of the video encoding device based on the intelligent digital retina of the present application, which can be used to execute the embodiment of the video encoding method based on the intelligent digital retina of the present application. For details not disclosed in the embodiment of the video encoding device based on the intelligent digital retina of the present application, please refer to the embodiment of the video encoding method based on the intelligent digital retina of the present application.

[0098] See also Figure 7 , which shows a schematic diagram of the structure of a video encoding device based on an intelligent digital retina provided by an exemplary embodiment of the present invention. The video encoding device based on the intelligent digital retina can be implemented as all or part of the terminal through software, hardware or a combination of both. The video encoding device based on the intelligent digital retina includes an acquisition module 702, a first image processing module 704, a calculation module 706 and an encoding module 708.

[0099] Specifically, the acquisition module 702 is used to acquire the original video image based on the collected video data;

[0100] A first image processing module 704 is used to input the original video image acquired by the acquisition module 702 into a first preset depth model for image processing, and obtain a corresponding feature sequence for image retrieval, a first feature region image of a first feature region, position information of the first feature region, and a first predicted image with a first preset resolution, wherein the feature sequence includes a feature sequence of a focus feature region;

[0101] A calculation module 706, configured to calculate a residual value between the first feature region image obtained by the image processing module and the first prediction image with a first preset resolution to obtain a first prediction residual;

[0102] The encoding module 708 is used to perform entropy encoding on the first prediction residual obtained by the calculation module 706 to obtain a corresponding first feature enhancement stream; and multiplex the first basic stream and the first feature enhancement stream to obtain an encoded video image.

[0103] Optionally, the encoding module 708 is specifically used for:

[0104] Downsampling the original video image to obtain a processed video image with a second preset resolution;

[0105] Performing video encoding on the processed video image to obtain a first basic stream;

[0106] The first basic stream and the first feature enhancement stream are multiplexed to obtain an encoded video image.

[0107] The device also includes:

[0108] Read module (in Figure 7 (not shown) for reading the position information and the feature area image of the feature area from the first frame buffer;

[0109] The encoding module 708 is also used to synchronously encode the position information and the feature area image read by the reading module.

[0110] It should be noted that the video encoding device based on the intelligent digital retina provided in the above embodiment only uses the division of the above functional units as an example when executing the video encoding device method based on the intelligent digital retina. In actual applications, the above functions can be assigned to different functional units as needed, that is, the internal structure of the device is divided into different functional units to complete all or part of the functions described above. In addition, the video encoding device based on the intelligent digital retina provided in the above embodiment and the video encoding method embodiment based on the intelligent digital retina belong to the same concept. The implementation process thereof is detailed in the video encoding method embodiment based on the intelligent digital retina, which will not be repeated here.

[0111] In an embodiment of the present application, an acquisition module is used to acquire an original video image based on collected video data; a first image processing module is used to input the original video image acquired by the acquisition module into a first preset depth model for image processing to obtain a corresponding feature sequence for image retrieval, a first feature area image of a first feature area, position information of the first feature area, and a first predicted image with a first preset resolution, wherein the feature sequence includes a feature sequence of a feature area of ​​interest; a calculation module is used to calculate a residual value of the first feature area image obtained by the image processing module and the first predicted image with a first preset resolution to obtain a first predicted residual; and an encoding module is used to perform entropy encoding on the first predicted residual obtained by the calculation module to obtain a corresponding first feature enhancement stream; and the first basic stream and the first feature enhancement stream are multiplexed to obtain an encoded video image. The video encoding method based on intelligent digital retina provided in the embodiment of the present application processes the video image by introducing a first preset depth model. Since the first preset depth model can extract image features, accurately determine the feature area of ​​the video image, and perform entropy encoding based on the first prediction residual to obtain a first feature enhancement stream, and obtain the encoded video image based on the first feature enhancement stream and the first basic stream; in this way, the encoded video image has better picture quality; in addition, the first preset depth model can also optimize the compression performance and playback performance of the video.

[0112] The embodiment of the present application also provides a video decoding device based on an intelligent digital retina, and the decoding device includes a reading and generating module, a second image processing module and a decoding module.

[0113] Specifically, the reading and generating module is used to respond to a playback instruction of a user to play back a target video, and the second frame buffer reads the second characteristic region of the target video image and generates a corresponding second characteristic region image;

[0114] A second image processing module, used for inputting the second feature area image obtained by the reading and generating module into a second preset depth model for image processing to obtain a corresponding third feature area image with a third preset resolution;

[0115] A decoding module is used to entropy decode a second feature enhancement stream corresponding to a second feature area to obtain a corresponding second prediction residual; and perform image synthesis based on the second prediction residual and a third feature area image to generate a first decoded target video image with a third preset resolution.

[0116] Optionally, the device further comprises:

[0117] An optimization module, configured to optimize the second preset depth model to obtain an optimized depth model, wherein the optimized depth model includes a third preset depth model for performing image optimization processing on a non-feature area of ​​a target video image;

[0118] The decoding module is also used to: decode the target video image based on the optimized depth model obtained by the optimization module to generate a second decoded target video image with a third preset resolution.

[0119] Optionally, the decoding module is specifically used for:

[0120] The second frame buffer generates a non-feature area image based on the second basic stream;

[0121] Inputting the non-feature area image into a third preset depth model for image processing to obtain a first target video image of the non-feature area having a third preset resolution;

[0122] Acquire a second target video image having a feature area of ​​the target video image and a third preset resolution;

[0123] The first target video image and the second target video image are subjected to image synthesis processing to generate a second decoded target video image having a third preset resolution.

[0124] like Figure 8 As shown, this embodiment provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement the method steps described above.

[0125] An embodiment of the present application provides a storage medium storing computer-readable instructions, on which a computer program is stored. The program is executed by a processor to implement the method steps described above.

[0126] Reference below Figure 8 , which shows a schematic diagram of the structure of an electronic device suitable for implementing the embodiments of the present application. The terminal device in the embodiments of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 8 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0127] like Figure 8As shown, the electronic device may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 801, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 802 or a program loaded from a storage device 808 to a random access memory (RAM) 803. Various programs and data required for the operation of the electronic device are also stored in the RAM 803. The processing device 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0128] Typically, the following devices may be connected to the I / O interface 805: input devices 806 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices 807 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 808 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 809. The communication device 809 may allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Although Figure 8 An electronic device having various devices is shown, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0129] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 809, or installed from the storage device 808, or installed from the ROM 802. When the computer program is executed by the processing device 801, the above-mentioned functions defined in the method of the embodiment of the present application are executed.

[0130] It should be noted that the computer-readable medium mentioned above in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. Computer readable signal media may also be any computer readable medium other than computer readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0131] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0132] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0133] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0134] The units involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of a unit does not, in some cases, constitute a limitation on the unit itself.

[0135] The above disclosure is only the preferred embodiment of the present application, which certainly cannot be used to limit the scope of rights of the present application. Therefore, equivalent changes made according to the claims of the present application are still within the scope covered by the present application.

Claims

1. A video encoding method based on intelligent digital retina, characterized in that: The method comprises: Acquire an original video image based on the collected video data; Inputting the original video image into a first preset depth model for image processing to obtain a corresponding feature sequence for image retrieval, a first feature region image of a first feature region, position information of the first feature region, and a first predicted image with a first preset resolution, wherein the feature sequence includes a feature sequence of a feature region of interest; the first preset resolution is a high resolution; Calculating a residual value between the first feature region image and the first predicted image with a first preset resolution to obtain a first prediction residual; Performing entropy coding on the first prediction residual to obtain a corresponding first feature enhancement stream; and multiplexing the first basic stream and the first feature enhancement stream to obtain an encoded video image; wherein the multiplexing of the first basic stream and the first feature enhancement stream to obtain an encoded video image includes: downsampling the original video image to obtain a processed video image with a second preset resolution; the second preset resolution is a low resolution; performing video coding on the processed video image to obtain the first basic stream; multiplexing the first basic stream and the first feature enhancement stream to obtain the encoded video image; Reading the position information of the feature area and the feature area image from a first frame buffer; The position information and the characteristic region image are synchronously encoded.

2. A video decoding method based on intelligent digital retina, characterized in that: The method comprises: In response to a playback instruction from a user to play back a target video, the second frame buffer reads a second feature region of the target video image and generates a corresponding second feature region image; Inputting the second feature region image into a second preset depth model for image processing to obtain a corresponding third feature region image with a third preset resolution; wherein the second preset depth model corresponds to the playback instruction; and the third preset resolution is a high resolution; Performing entropy decoding on a second feature enhancement stream corresponding to the second feature area to obtain a corresponding second prediction residual; and performing image synthesis based on the second prediction residual, a non-feature area image of a target video image, and the third feature area image to generate a first decoded target video image having the third preset resolution; Optimizing the second preset depth model to obtain an optimized depth model, wherein the optimized depth model includes a third preset depth model that performs image optimization processing on a non-feature area of ​​a target video image; Based on the optimized depth model, the target video image is decoded and processed to generate a second decoded target video image with the third preset resolution, including: the second frame buffer generates a non-feature area image based on the second basic stream; the non-feature area image is input into the third preset depth model for image processing to obtain a first target video image of the non-feature area with the third preset resolution; the feature area of ​​the target video image is obtained, and the second target video image with the third preset resolution is obtained; the first target video image and the second target video image are subjected to image synthesis processing to generate a second decoded target video image with the third preset resolution.

3. A video encoding device based on intelligent digital retina, characterized in that: The device comprises: An acquisition module, used to acquire an original video image based on the collected video data; A first image processing module, used for inputting the original video image acquired by the acquisition module into a first preset depth model for image processing, to obtain a corresponding feature sequence for image retrieval, a first feature region image of a first feature region, position information of the first feature region, and a first predicted image with a first preset resolution, wherein the feature sequence includes a feature sequence of a feature region of interest; the first preset resolution is a high resolution; a calculation module, configured to calculate a residual value between the first feature region image obtained by the image processing module and the first prediction image with a first preset resolution to obtain a first prediction residual; A coding module, configured to perform entropy coding on the first prediction residual obtained by the calculation module to obtain a corresponding first feature enhancement stream; and multiplex the first basic stream and the first feature enhancement stream to obtain a coded video image; wherein the multiplexing of the first basic stream and the first feature enhancement stream to obtain the coded video image comprises: downsampling the original video image to obtain a processed video image with a second preset resolution; the second preset resolution is a low resolution; performing video coding on the processed video image to obtain the first basic stream; and multiplexing the first basic stream and the first feature enhancement stream to obtain the coded video image; Reading the position information of the feature area and the feature area image from a first frame buffer; The position information and the characteristic region image are synchronously encoded.

4. A video decoding device based on intelligent digital retina, characterized in that: The device comprises: A reading and generating module, configured to respond to a playback instruction of a user to play back a target video, and to cause a second frame buffer to read a second characteristic region of a target video image and generate a corresponding second characteristic region image; A second image processing module, used for inputting the second feature area image obtained by the reading and generating module into a second preset depth model for image processing to obtain a corresponding third feature area image with a third preset resolution; wherein the second preset depth model corresponds to the playback instruction; and the third preset resolution is a high resolution; A decoding module, configured to perform entropy decoding on a second feature enhancement stream corresponding to the second feature area to obtain a corresponding second prediction residual; and perform image synthesis based on the second prediction residual, a non-feature area image of a target video image, and the third feature area image to generate a first decoded target video image having the third preset resolution; An optimization module, configured to optimize the second preset depth model to obtain an optimized depth model, wherein the optimized depth model includes a third preset depth model for performing image optimization processing on a non-feature area of ​​a target video image; A decoding generation module is used to decode the target video image based on the optimized depth model to generate a second decoded target video image with the third preset resolution; a decoding generation module is specifically used for the second frame buffer to generate a non-feature area image based on the second basic stream; the non-feature area image is input into the third preset depth model for image processing to obtain a first target video image of the non-feature area with the third preset resolution; the feature area of ​​the target video image is obtained and the second target video image with the third preset resolution is obtained; the first target video image and the second target video image are subjected to image synthesis processing to generate a second decoded target video image with the third preset resolution.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: The processor runs the computer program to implement the method according to any one of claims 1-2.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the method according to any one of claims 1 to 2.