CN-122023215-A - Image optimization enhancement method based on artificial intelligence

CN122023215ACN 122023215 ACN122023215 ACN 122023215ACN-122023215-A

Abstract

The invention relates to the technical field of image enhancement and discloses an image optimization enhancement method based on artificial intelligence, which comprises the steps of carrying out multidimensional image feature analysis on an image to be optimized to obtain key quality index data; the method comprises the steps of matching hierarchical optimization parameters corresponding to an image to be optimized from a preset parameter mapping rule, carrying out cooperative adjustment on contrast and brightness of the image to be optimized to obtain a preliminary optimized image, carrying out multi-scale fringe sensing detection on the preliminary optimized image to divide a detail strengthening area and a smooth keeping area to obtain area division mask data, carrying out differential strengthening treatment on the detail strengthening area and the smooth keeping area in the preliminary optimized image to obtain optimized image component data, and carrying out weighted synthesis on the optimized image component data and the preliminary optimized image to obtain a final optimized image.

Inventors

WANG HANFANG
CHAI YI
ZHANG HUI
San Tiancheng

Assignees

武汉软件工程职业学院(武汉开放大学)

Dates

Publication Date: 20260512
Application Date: 20260205

Claims (10)

1. An artificial intelligence-based image optimization enhancement method, the method comprising: s1, carrying out multidimensional image feature analysis on an image to be optimized to obtain key quality index data of the image to be optimized; S2, matching hierarchical optimization parameters corresponding to the image to be optimized from a preset parameter mapping rule according to the key quality index data; S3, based on the hierarchical optimization parameters, performing cooperative adjustment on the contrast and brightness of the image to be optimized to obtain a preliminary optimized image of the image to be optimized; S4, performing multi-scale fringe sensing detection on the preliminary optimized image to divide a detail strengthening area and a smooth keeping area in the preliminary optimized image, and obtaining area division mask data of the preliminary optimized image; s5, based on the region division mask data, performing differential enhancement processing on a detail enhancement region and a smooth keeping region in the preliminary optimized image to obtain optimized image component data of the preliminary optimized image; And S6, carrying out weighted synthesis on the optimized image component data and the preliminary optimized image to obtain a final optimized image of the image to be optimized.
2. The image optimization and enhancement method based on artificial intelligence as claimed in claim 1, wherein the performing multi-dimensional image feature analysis on the image to be optimized to obtain key quality index data of the image to be optimized includes: Acquiring original pixel data of an image to be optimized; Extracting color feature distribution information, texture feature statistical information and definition feature description information in the original pixel data; Carrying out standardization processing on the color feature distribution information, the texture feature statistical information and the definition feature description information to obtain a feature vector set of the image to be optimized; And inputting the feature vector set into a preset feature quantization rule to obtain the key quality index data of the image to be optimized.
3. The method for enhancing image optimization based on artificial intelligence according to claim 1, wherein the matching the hierarchical optimization parameters corresponding to the image to be optimized from the preset parameter mapping rule according to the key quality index data comprises: Matching degree comparison is carried out on the color balance degree index, the texture complexity index and the image ambiguity index in the key quality index data and a grade threshold value in a preset parameter mapping rule, so that the comprehensive quality grade of the image to be optimized is obtained; screening a basic optimization parameter set corresponding to the image to be optimized from the preset parameter mapping rule according to the comprehensive quality grade; And based on the relative relation between the texture complexity index and the image ambiguity index, cooperatively adjusting the basic optimization parameter set to obtain the hierarchical optimization parameters of the image to be optimized.
4. The method for enhancing the image optimization based on the artificial intelligence according to claim 1, wherein the collaborative adjustment of the contrast and brightness of the image to be optimized based on the hierarchical optimization parameters, to obtain a preliminary optimized image of the image to be optimized, comprises: separating a contrast adjustment coefficient and a brightness reference value from the hierarchical optimization parameters; according to the contrast adjustment coefficient, carrying out transformation adjustment on the pixel gray values in the image to be optimized to obtain an intermediate adjustment image of the image to be optimized; Counting the overall brightness average value of the intermediate adjustment image, and performing global translation compensation on all pixel values of the intermediate adjustment image according to the difference value between the brightness reference value and the overall brightness average value; And performing brightness clipping on the image subjected to global translation compensation, and restricting all pixel values to be within an effective numerical range to obtain a preliminary optimized image of the image to be optimized.
5. The method for image optimization and enhancement based on artificial intelligence according to claim 1, wherein the performing multi-scale fringe sensing on the preliminary optimized image to divide a detail enhancement area and a smooth preservation area in the preliminary optimized image to obtain the area division mask data of the preliminary optimized image includes: Performing edge detection on the image data with different sizes in the preliminary optimized image to obtain an initial edge response diagram of the preliminary optimized image; The edge contour information in the initial edge response graph is subjected to non-maximum value inhibition processing to obtain an edge intensity graph of the preliminary optimized image; Carrying out weighted fusion on pixel points in the edge intensity image to obtain a comprehensive edge response image of the preliminary optimized image; Performing self-adaptive threshold segmentation on the comprehensive edge response graph, marking a pixel region with a response value higher than a dynamic threshold as a detail strengthening region, and marking a pixel region with a response value lower than the dynamic threshold as a smooth maintaining region; And counting marking results of the detail strengthening area and the smooth keeping area to obtain area division mask data of the preliminary optimized image.
6. The artificial intelligence based image optimization enhancement method according to claim 5, wherein the performing edge detection on the image data with different sizes in the preliminary optimized image to obtain an initial edge response map of the preliminary optimized image of the image to be optimized includes: taking a gradient operator of the detection scale configuration in the image to be optimized, which is used for capturing a significant edge and corresponds to capturing a fine texture, as the gradient operator of the detection scale configuration; Performing two-dimensional convolution operation on the preliminary optimized image based on the gradient operator to obtain a horizontal gradient component diagram and a vertical gradient component diagram of the detection scale configuration; determining an edge intensity value of the preliminary optimized image according to the horizontal gradient component and the vertical gradient component; and integrating the pixel points corresponding to the edge intensity values to obtain an initial edge response diagram of the image to be optimized.
7. The method for image optimization and enhancement based on artificial intelligence according to claim 5, wherein the step of performing weighted fusion on the pixels in the edge intensity image to obtain the comprehensive edge response image of the preliminary optimized image comprises the steps of: Different dimension data caused by the dimension difference in the edge intensity image are removed, and an adjusted edge intensity image of the preliminary optimized image is obtained; Determining fusion weight coefficients of different scale images according to the contribution degrees of the different scale images in the adjusted edge intensity image in the edge detection task; Based on the fusion weight coefficient, carrying out accumulated response on the images with different scales to obtain a comprehensive response value of the preliminary optimized image; And traversing and integrating the corresponding pixel points of the comprehensive response value to obtain a comprehensive edge response diagram of the preliminary optimized image.
8. The image optimization enhancement method based on artificial intelligence according to claim 1, wherein the performing differential enhancement processing on a detail enhancement region and a smooth preservation region in the preliminary optimized image based on the region division mask data to obtain optimized image component data of the preliminary optimized image includes: Extracting pixels in the detail strengthening area and the smooth keeping area from the preliminary optimized image according to the marked pixel positions in the area division mask data to form a first image block set and a second image block set of the preliminary optimized image; Local contrast enhancement is carried out on the first image block set, and edge sharpening operation is carried out on the enhanced first image block set, so that detail enhancement image components of the preliminary optimization image are obtained; removing noise in the second image block set to obtain a smooth optimized image component of the preliminary optimized image; Recombining pixels in the detail enhanced image component and pixels in the smooth optimized image component according to the original position information recorded by the regional division mask data to obtain an intermediate result image of the preliminary optimized image; And performing color space consistency check on the intermediate result image to obtain optimized image component data of the preliminary optimized image.
9. The image optimization and enhancement method based on artificial intelligence according to claim 8, wherein the performing local contrast enhancement on the first image block set and performing edge sharpening operation on the enhanced first image block set to obtain a detail enhancement image component of the preliminary optimization image includes: Dividing the first image block set into a local sub-region set; Performing histogram equalization on the local subarea sets, and performing weighted average fusion on the equalized local subarea sets based on an overlapping relation among the local subarea sets to obtain a contrast enhancement image block of the preliminary optimized image; adding a preset gain coefficient to the high-frequency edge component in the contrast enhancement image block to obtain a contrast superposition image block of the preliminary optimized image; And cutting out the overflow value in the contrast superposition image block to obtain the detail enhancement image component of the preliminary optimization image.
10. The artificial intelligence based image optimization enhancement method of claim 1, wherein the weighting and synthesizing the optimized image component data and the preliminary optimized image to obtain a final optimized image of the image to be optimized comprises: determining an optimized component weight corresponding to the optimized image component data and a preliminary image weight corresponding to the preliminary optimized image according to a synthesis strategy defined in the hierarchical optimized parameters; Performing weighted scaling on pixel values in the optimized image component data and the preliminary optimized image based on the optimized component weight and the preliminary image weight to obtain an optimized weighted component of the optimized image component data and a preliminary weighted component of the preliminary optimized image; Fusing the optimized weighted component and the preliminary weighted component aiming at the pixel positions in the optimized image component data and the preliminary optimized image to obtain a synthesized pixel value of the preliminary optimized image; and integrating the synthesized pixel values to obtain an initial synthesized image of the image to be optimized, and carrying out numerical range constraint on the initial synthesized image to obtain a final optimized image of the image to be optimized.

Description

Image optimization enhancement method based on artificial intelligence Technical Field The invention relates to the technical field of image enhancement, in particular to an image optimization enhancement method based on artificial intelligence. Background In the technical field of image enhancement, a traditional image optimization method often adopts a single parameter adjustment strategy, and lacks of accurate adaptation to multi-dimensional characteristics of an image. The method generally depends on a fixed contrast and brightness adjustment formula, and key quality characteristics such as color distribution balance, texture complexity and definition difference of the image are not fully considered, so that the optimization process is blind, an adaptive optimization scheme is difficult to form for images with different quality levels, and the problems of image detail blurring, color distortion or smooth region noise amplification and the like cannot be fundamentally solved. In the prior art, an effective partition enhancement mechanism is lacked when the difference of the image areas is processed, and a global unified enhancement algorithm is mostly adopted, so that detail enhancement areas and smooth keeping areas in the image cannot be distinguished. The single processing mode is easy to cause contradiction of insufficient enhancement of detail areas and excessive sharpening of smooth areas, not only affects the overall effect of image optimization, but also reduces processing efficiency due to increased invalid calculation, and is difficult to meet the dual requirements of image enhancement precision and speed in practical application, so that the efficient landing of the image enhancement technology in various scenes is restricted, and therefore, how to improve the image optimization enhancement efficiency based on artificial intelligence becomes a problem to be solved urgently. Disclosure of Invention The invention provides an image optimization enhancement method based on artificial intelligence, which aims to solve the problems in the background technology. In order to achieve the above object, the present invention provides an image optimization enhancing method based on artificial intelligence, comprising: s1, carrying out multidimensional image feature analysis on an image to be optimized to obtain key quality index data of the image to be optimized; S2, matching hierarchical optimization parameters corresponding to the image to be optimized from a preset parameter mapping rule according to the key quality index data; S3, based on the hierarchical optimization parameters, performing cooperative adjustment on the contrast and brightness of the image to be optimized to obtain a preliminary optimized image of the image to be optimized; S4, performing multi-scale fringe sensing detection on the preliminary optimized image to divide a detail strengthening area and a smooth keeping area in the preliminary optimized image, and obtaining area division mask data of the preliminary optimized image; s5, based on the region division mask data, performing differential enhancement processing on a detail enhancement region and a smooth keeping region in the preliminary optimized image to obtain optimized image component data of the preliminary optimized image; And S6, carrying out weighted synthesis on the optimized image component data and the preliminary optimized image to obtain a final optimized image of the image to be optimized. In a preferred embodiment, the performing multidimensional image feature analysis on the image to be optimized to obtain key quality index data of the image to be optimized includes: Acquiring original pixel data of an image to be optimized; Extracting color feature distribution information, texture feature statistical information and definition feature description information in the original pixel data; Carrying out standardization processing on the color feature distribution information, the texture feature statistical information and the definition feature description information to obtain a feature vector set of the image to be optimized; And inputting the feature vector set into a preset feature quantization rule to obtain the key quality index data of the image to be optimized. In a preferred embodiment, the matching, according to the key quality indicator data, the hierarchical optimization parameter corresponding to the image to be optimized from a preset parameter mapping rule includes: Matching degree comparison is carried out on the color balance degree index, the texture complexity index and the image ambiguity index in the key quality index data and a grade threshold value in a preset parameter mapping rule, so that the comprehensive quality grade of the image to be optimized is obtained; screening a basic optimization parameter set corresponding to the image to be optimized from the preset parameter mapping rule according to the comprehensive quality grade; And b