Application picture enhancement system based on neural network
Through a neural network-based picture enhancement system, BP neural network analyzes and predicts picture content, selects suitable enhancement algorithms to process different pictures, solving the problem of poor enhancement effect in the existing technology, and realizing intelligent prediction and customized pictures Enhanced.
Patent Information
- Application Number
- CN202411179758.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-12-15
- Filing Date
- 2024-08-27
- Publication Date
- 2024-11-29
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the prior art, when different pictures taken by the same camera software are processed using the same enhancement mode, the results are very different, making it difficult to choose the best enhancement algorithm suitable for each picture.
Design a neural network-based application image enhancement system, through content input, continuous operation, prediction operation and dynamic enhancement mechanism, use BP neural network to perform dynamic threshold white balance, artifact removal and Lanczos interpolation processing on pictures, and analyze the picture Multiple content data and noise types are selected to select the most contrast enhancement algorithm for processing.
Intelligent prediction and customized enhancement processing for different pictures are realized to ensure that each picture obtains the appropriate optimal enhancement effect, and solves the problem of poor enhancement effect in the existing technology.
Abstract
Description
Technical Field
[0001] The present invention relates to the field of image processing, and in particular to an application image enhancement system based on a neural network. Background Art
[0002] Camera software is a software that everyone uses. It appears in various electronic devices and mobile terminals. Camera software can help users take the beautiful photos they want in minutes. There is even a concept of acoustic camera (also known as sound imager, acoustic camera, etc.). It is a device used to analyze, locate and visualize sound sources. It can not only let people know where the sound comes from, but also further show the characteristics of the sound.
[0003] However, when people use camera software to take pictures on various electronic devices and mobile terminals, although the various setting parameters of the camera software are the same, the content and characteristics of different pictures taken by the same camera software are still very different. Using the same enhancement mode to perform the same enhancement processing on different pictures taken by the same camera software may result in unsatisfactory results. How to select the best enhancement algorithm suitable for different pictures taken by the same camera software is a technical problem that needs to be solved in the prior art. Summary of the invention
[0004] To this end, the present invention proposes an application picture enhancement system based on a neural network, the system comprising: A content input mechanism, arranged in the mobile terminal, for receiving a picture to be processed selected by a user of the mobile terminal through a selection mechanism of the mobile terminal, wherein the picture to be processed is selected from various pictures stored in a camera application of the mobile terminal; A continuous operation mechanism, which is arranged in the mobile terminal and connected to the content input mechanism, comprises a dynamic processing device, an artifact removal device and a neighborhood interpolation device. The dynamic processing device is connected to the content input mechanism and is used to perform white balance processing based on a dynamic threshold on the received image to be processed, so as to obtain and output a corresponding dynamic processing image. The artifact removal device is connected to the dynamic processing device and is used to perform artifact removal processing on the received dynamic processing image, so as to obtain and output a corresponding artifact removal image. The neighborhood interpolation device is connected to the artifact removal device and is used to perform Lanczos interpolation processing of an 8-pixel×8-pixel neighborhood on the received artifact removal image, so as to obtain and output a corresponding neighborhood interpolation image. a prediction operation mechanism connected to the continuous operation mechanism, used for acquiring the neighborhood interpolation image, analyzing each pixel value corresponding to each pixel point of the neighborhood interpolation image, and determining multiple content data of the neighborhood interpolation image, wherein the multiple content data of the neighborhood interpolation image are the number of pixel rows, the number of pixel columns and the total number of main noise types of the neighborhood interpolation image, performing each learning process of the BP neural network for the number of times reversely associated with the clarity of the neighborhood interpolation image to obtain the corresponding BP neural network model, inputting the multiple content data of the neighborhood interpolation image, each pixel value corresponding to each pixel point of the neighborhood interpolation image and the code value representation of the current enhancement algorithm into the BP neural network model at the same time, and running the BP neural network model to obtain the contrast of the picture obtained after the current enhancement algorithm is executed on the neighborhood interpolation image, wherein the main noise type is the noise type whose noise amplitude exceeds the set amplitude limit; The dynamic enhancement mechanism is connected to the prediction operation mechanism and is used to select the enhancement algorithm with the highest contrast of the corresponding picture from various enhancement algorithms and apply it to the neighborhood interpolation image to realize the enhancement processing of the neighborhood interpolation image and obtain the corresponding dynamic enhanced image.
[0005] According to a neural network-based application picture enhancement system proposed in the present invention, for a customized optimized image, i.e., a neighborhood interpolation image, corresponding to a picture taken by the same camera software, multiple content data of the neighborhood interpolation image, pixel values corresponding to each pixel point of the neighborhood interpolation image, and a code numerical representation of the current enhancement algorithm are simultaneously input into a BP neural network model, and the BP neural network model is run to obtain the output of the contrast of the picture obtained after the current enhancement algorithm is executed on the neighborhood interpolation image, thereby realizing intelligent prediction of the enhancement effects after being processed by various enhancement algorithms, and providing valuable reference information for the subsequent selection of the type of enhancement algorithm. DETAILED DESCRIPTION
[0006] The implementation scheme of the neural network-based application picture enhancement system of the present invention will be described in detail below.
[0007] AImplementation Plan According to embodiment A of the present invention, a neural network-based application picture enhancement system is shown, the system comprising: A content input mechanism, arranged in the mobile terminal, for receiving a picture to be processed selected by a user of the mobile terminal through a selection mechanism of the mobile terminal, wherein the picture to be processed is selected from various pictures stored in a camera application of the mobile terminal; Specifically, the content input mechanism is arranged in the mobile terminal, and is used to receive the to-be-processed picture selected by the user of the mobile terminal through the selection mechanism of the mobile terminal, wherein the to-be-processed picture is from each picture stored in the camera application of the mobile terminal, and includes: the mobile terminal adopts a built-in storage device to store each picture to be stored in the camera application; A continuous operation mechanism, which is arranged in the mobile terminal and connected to the content input mechanism, comprises a dynamic processing device, an artifact removal device and a neighborhood interpolation device. The dynamic processing device is connected to the content input mechanism and is used to perform white balance processing based on a dynamic threshold on the received image to be processed, so as to obtain and output a corresponding dynamic processing image. The artifact removal device is connected to the dynamic processing device and is used to perform artifact removal processing on the received dynamic processing image, so as to obtain and output a corresponding artifact removal image. The neighborhood interpolation device is connected to the artifact removal device and is used to perform Lanczos interpolation processing of an 8-pixel×8-pixel neighborhood on the received artifact removal image, so as to obtain and output a corresponding neighborhood interpolation image. a prediction operation mechanism connected to the continuous operation mechanism, used for acquiring the neighborhood interpolation image, analyzing each pixel value corresponding to each pixel point of the neighborhood interpolation image, and determining multiple content data of the neighborhood interpolation image, wherein the multiple content data of the neighborhood interpolation image are the number of pixel rows, the number of pixel columns and the total number of main noise types of the neighborhood interpolation image, performing each learning process of the BP neural network for the number of times reversely associated with the clarity of the neighborhood interpolation image to obtain the corresponding BP neural network model, inputting the multiple content data of the neighborhood interpolation image, each pixel value corresponding to each pixel point of the neighborhood interpolation image and the code value representation of the current enhancement algorithm into the BP neural network model at the same time, and running the BP neural network model to obtain the contrast of the picture obtained after the current enhancement algorithm is executed on the neighborhood interpolation image, wherein the main noise type is the noise type whose noise amplitude exceeds the set amplitude limit; A dynamic enhancement mechanism, connected to the prediction operation mechanism, for selecting an enhancement algorithm with the highest contrast of the corresponding picture from various enhancement algorithms and applying it to the neighborhood interpolation image, so as to implement enhancement processing on the neighborhood interpolation image and obtain a corresponding dynamic enhanced image; Among them, the picture to be processed selected by the user of the receiving mobile terminal through the selection mechanism of the mobile terminal, the picture to be processed comes from the pictures stored in the camera application of the mobile terminal, including: the pictures stored in the camera application of the mobile terminal have different picture numbers.
[0008] B Implementation Plan Embodiment B according to the present invention shows an application picture enhancement system based on a neural network.
[0009] The neural network-based application picture enhancement system shown in the B embodiment of the present invention may include: A content input mechanism, arranged in the mobile terminal, for receiving a picture to be processed selected by a user of the mobile terminal through a selection mechanism of the mobile terminal, wherein the picture to be processed is selected from various pictures stored in a camera application of the mobile terminal; A continuous operation mechanism, which is arranged in the mobile terminal and connected to the content input mechanism, comprises a dynamic processing device, an artifact removal device and a neighborhood interpolation device. The dynamic processing device is connected to the content input mechanism and is used to perform white balance processing based on a dynamic threshold on the received image to be processed, so as to obtain and output a corresponding dynamic processing image. The artifact removal device is connected to the dynamic processing device and is used to perform artifact removal processing on the received dynamic processing image, so as to obtain and output a corresponding artifact removal image. The neighborhood interpolation device is connected to the artifact removal device and is used to perform Lanczos interpolation processing of an 8-pixel×8-pixel neighborhood on the received artifact removal image, so as to obtain and output a corresponding neighborhood interpolation image. a prediction operation mechanism connected to the continuous operation mechanism, used for acquiring the neighborhood interpolation image, analyzing each pixel value corresponding to each pixel point of the neighborhood interpolation image, and determining multiple content data of the neighborhood interpolation image, wherein the multiple content data of the neighborhood interpolation image are the number of pixel rows, the number of pixel columns and the total number of main noise types of the neighborhood interpolation image, performing each learning process of the BP neural network for the number of times reversely associated with the clarity of the neighborhood interpolation image to obtain the corresponding BP neural network model, inputting the multiple content data of the neighborhood interpolation image, each pixel value corresponding to each pixel point of the neighborhood interpolation image and the code value representation of the current enhancement algorithm into the BP neural network model at the same time, and running the BP neural network model to obtain the contrast of the picture obtained after the current enhancement algorithm is executed on the neighborhood interpolation image, wherein the main noise type is the noise type whose noise amplitude exceeds the set amplitude limit; A dynamic enhancement mechanism, connected to the prediction operation mechanism, for selecting an enhancement algorithm with the highest contrast of the corresponding picture from various enhancement algorithms and applying it to the neighborhood interpolation image, so as to implement enhancement processing on the neighborhood interpolation image and obtain a corresponding dynamic enhanced image; a synchronous driving device, which is arranged near the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device and is connected to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively; Among them, the synchronous driving device is arranged near the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device and is respectively connected to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device, including: the synchronous driving device is used to respectively realize the synchronous driving control of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device in pairs.
[0010] C Implementation Plan Embodiment C of the present invention shows an application picture enhancement system based on a neural network.
[0011] The neural network-based application picture enhancement system shown in the C embodiment of the present invention may include: A content input mechanism, arranged in the mobile terminal, for receiving a picture to be processed selected by a user of the mobile terminal through a selection mechanism of the mobile terminal, wherein the picture to be processed is selected from various pictures stored in a camera application of the mobile terminal; A continuous operation mechanism, which is arranged in the mobile terminal and connected to the content input mechanism, comprises a dynamic processing device, an artifact removal device and a neighborhood interpolation device. The dynamic processing device is connected to the content input mechanism and is used to perform white balance processing based on a dynamic threshold on the received image to be processed, so as to obtain and output a corresponding dynamic processing image. The artifact removal device is connected to the dynamic processing device and is used to perform artifact removal processing on the received dynamic processing image, so as to obtain and output a corresponding artifact removal image. The neighborhood interpolation device is connected to the artifact removal device and is used to perform Lanczos interpolation processing of an 8-pixel×8-pixel neighborhood on the received artifact removal image, so as to obtain and output a corresponding neighborhood interpolation image. a prediction operation mechanism connected to the continuous operation mechanism, used for acquiring the neighborhood interpolation image, analyzing each pixel value corresponding to each pixel point of the neighborhood interpolation image, and determining multiple content data of the neighborhood interpolation image, wherein the multiple content data of the neighborhood interpolation image are the number of pixel rows, the number of pixel columns and the total number of main noise types of the neighborhood interpolation image, performing each learning process of the BP neural network for the number of times reversely associated with the clarity of the neighborhood interpolation image to obtain the corresponding BP neural network model, inputting the multiple content data of the neighborhood interpolation image, each pixel value corresponding to each pixel point of the neighborhood interpolation image and the code value representation of the current enhancement algorithm into the BP neural network model at the same time, and running the BP neural network model to obtain the contrast of the picture obtained after the current enhancement algorithm is executed on the neighborhood interpolation image, wherein the main noise type is the noise type whose noise amplitude exceeds the set amplitude limit; A dynamic enhancement mechanism, connected to the prediction operation mechanism, for selecting an enhancement algorithm with the highest contrast of the corresponding picture from various enhancement algorithms and applying it to the neighborhood interpolation image, so as to implement enhancement processing on the neighborhood interpolation image and obtain a corresponding dynamic enhanced image; A circuit supply device, which is arranged near the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device and is respectively connected to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device; Among them, the circuit supply device is arranged near the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device and is respectively connected to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device, including: the circuit supply device is used to provide the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device with the required working voltage values respectively.
[0012] Next, the specific structure of the neural network-based application picture enhancement system of the present invention will be further described.
[0013] In the neural network-based application picture enhancement system according to various embodiments of the present invention: An ASIC chip is used to perform image data processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively.
[0014] In the neural network-based application picture enhancement system according to various embodiments of the present invention: Using an ASIC chip to perform image data processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively includes: performing maximum value filtering processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively.
[0015] In the neural network-based application picture enhancement system according to various embodiments of the present invention: Using an ASIC chip to perform image data processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively includes: performing minimum value filtering processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively.
[0016] In the neural network-based application picture enhancement system according to various embodiments of the present invention: Using an ASIC chip to perform image data processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively includes: performing median filtering processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively.
[0017] And in the neural network-based application picture enhancement system according to various embodiments of the present invention: Using an ASIC chip to perform image data processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively includes: performing edge sharpening processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively.
[0018] In addition, in the neural network-based application picture enhancement system, receiving the picture to be processed selected by the user of the mobile terminal through the selection mechanism of the mobile terminal, the picture to be processed comes from the various pictures stored in the camera application of the mobile terminal and also includes: using different physical addresses to perform directional storage of the various pictures stored in the camera application of the mobile terminal.
[0019] It can be seen that the present invention has at least the following beneficial technical advances: Firstly: analyzing and selecting pixel values corresponding to pixel points of a neighborhood interpolation image obtained after targeted processing of the image, and determining multiple content data of the neighborhood interpolation image, wherein the multiple content data of the neighborhood interpolation image are the number of pixel rows, the number of pixel columns and the total number of main noise types of the neighborhood interpolation image; Secondly: the BP neural network is subjected to each learning process of the number of times reversely associated with the clarity of the neighborhood interpolation image to obtain the corresponding BP neural network model, thereby completing the structural customization of the artificial intelligence model; Again: multiple content data of the neighborhood interpolation image, each pixel value corresponding to each pixel point of the neighborhood interpolation image, and the code numerical representation of the current enhancement algorithm are simultaneously input into the BP neural network model, and the BP neural network model is run to obtain the output of the contrast of the picture obtained after the current enhancement algorithm is executed on the neighborhood interpolation image, thereby realizing intelligent prediction of the enhancement effects after various enhancement algorithms are processed separately, and providing valuable reference information for the subsequent selection of the type of enhancement algorithm.
[0020] In general, the neural network-based application picture enhancement system of the present invention aims to solve the technical problem in the prior art that different pictures taken by the same camera software have poor enhancement effects when the same enhancement processing algorithm is used. By being able to customize optimized images corresponding to pictures taken by the same camera software, namely neighborhood interpolation images, the contrast of different pictures obtained after different enhancement algorithms are executed on the neighborhood interpolation images can be intelligently analyzed, thereby realizing intelligent prediction of the enhancement effects after various enhancement algorithms are processed separately, and ensuring that different pictures can obtain the best enhancement algorithm suitable for them.
[0021] Although the above disclosure is about the preferred embodiment of the present invention, it should be understood that the above details are for illustration only. Those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention.
Claims
1. A neural network-based application picture enhancement system, characterized in that: The system comprises: A content input mechanism, arranged in the mobile terminal, for receiving a picture to be processed selected by a user of the mobile terminal through a selection mechanism of the mobile terminal, wherein the picture to be processed is selected from various pictures stored in a camera application of the mobile terminal; A continuous operation mechanism, which is arranged in the mobile terminal and connected to the content input mechanism, comprises a dynamic processing device, an artifact removal device and a neighborhood interpolation device. The dynamic processing device is connected to the content input mechanism and is used to perform white balance processing based on a dynamic threshold on the received image to be processed, so as to obtain and output a corresponding dynamic processing image. The artifact removal device is connected to the dynamic processing device and is used to perform artifact removal processing on the received dynamic processing image, so as to obtain and output a corresponding artifact removal image. The neighborhood interpolation device is connected to the artifact removal device and is used to perform Lanczos interpolation processing of an 8-pixel×8-pixel neighborhood on the received artifact removal image, so as to obtain and output a corresponding neighborhood interpolation image. a prediction operation mechanism connected to the continuous operation mechanism, used for acquiring the neighborhood interpolation image, analyzing each pixel value corresponding to each pixel point of the neighborhood interpolation image, and determining multiple content data of the neighborhood interpolation image, wherein the multiple content data of the neighborhood interpolation image are the number of pixel rows, the number of pixel columns and the total number of main noise types of the neighborhood interpolation image, performing each learning process of the BP neural network for the number of times reversely associated with the clarity of the neighborhood interpolation image to obtain the corresponding BP neural network model, inputting the multiple content data of the neighborhood interpolation image, each pixel value corresponding to each pixel point of the neighborhood interpolation image and the code value representation of the current enhancement algorithm into the BP neural network model at the same time, and running the BP neural network model to obtain the contrast of the picture obtained after the current enhancement algorithm is executed on the neighborhood interpolation image, wherein the main noise type is the noise type whose noise amplitude exceeds the set amplitude limit; The dynamic enhancement mechanism is connected to the prediction operation mechanism and is used to select the enhancement algorithm with the highest contrast of the corresponding picture from various enhancement algorithms and apply it to the neighborhood interpolation image to realize the enhancement processing of the neighborhood interpolation image and obtain the corresponding dynamic enhanced image.
2. The neural network-based application picture enhancement system according to claim 1, characterized in that: The picture to be processed selected by the user of the mobile terminal through the selection mechanism of the mobile terminal is received, and the picture to be processed is from each picture stored in the camera application of the mobile terminal, including: each picture stored in the camera application of the mobile terminal has a different picture number.
3. The neural network-based application picture enhancement system according to claim 2, characterized in that: The system further comprises: a synchronous driving device, which is arranged near the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device and is connected to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively; Among them, the synchronous driving device is arranged near the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device and is respectively connected to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device, including: the synchronous driving device is used to respectively realize the synchronous driving control of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device in pairs.
4. The neural network-based application picture enhancement system as claimed in claim 2, characterized in that: The system further comprises: A circuit supply device, which is arranged near the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device and is respectively connected to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device; Among them, the circuit supply device is arranged near the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device and is respectively connected to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device, including: the circuit supply device is used to provide the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device with the required working voltage values respectively.
5. The neural network-based application picture enhancement system according to any one of claims 2 to 4, characterized in that: An ASIC chip is used to perform image data processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively.
6. The neural network-based application picture enhancement system according to claim 5, characterized in that: Using an ASIC chip to perform image data processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively includes: performing maximum value filtering processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively.
7. The neural network-based application picture enhancement system according to claim 5, characterized in that: Using an ASIC chip to perform image data processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively includes: performing minimum value filtering processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively.
8. The neural network-based application picture enhancement system according to claim 5, characterized in that: Using an ASIC chip to perform image data processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively includes: performing median filtering processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively.
9. The neural network-based application picture enhancement system according to claim 5, characterized in that: Using an ASIC chip to perform image data processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively includes: performing edge sharpening processing on the output data of the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device to obtain the output processing data corresponding to the prediction operation mechanism, the dynamic processing device, the artifact removal device and the neighborhood interpolation device respectively.
Citation Information
Cited By
Entrance guard monitoring image dynamic processing system
CN119360488A
Image enhancement method and system based on image processing chip
CN119399050A
