CN-121982002-A - Image processing method and system for enhanced skin pressure injury monitoring

CN121982002ACN 121982002 ACN121982002 ACN 121982002ACN-121982002-A

Abstract

The invention discloses an image processing method and an image processing system for enhanced skin pressure injury monitoring, which relate to the technical field of image processing and comprise the following steps of performing self-adaptive partition processing on skin images; the method comprises the steps of extracting texture entropy, gradient distribution and brightness deviation characteristics of each local block, constructing a local micro-damage response graph, constructing an adaptive enhancement function, performing enhancement processing, executing interactive verification of multiple sets of enhanced images, calculating consistency scores, executing interactive enhancement processing, executing offset residual analysis, constructing a local enhanced reliability graph, triggering secondary adaptive enhancement processing, performing interactive fusion on a first enhanced image and a second enhanced image, and outputting a third enhanced image. The technical problems that the conventional image enhancement in the prior art cannot adapt to the local heterogeneity of skin images, the early weak damage features are difficult to extract effectively and the reliability is low are solved, and the technical effects of realizing the accurate enhancement processing of the local micro damage features so as to ensure the reliability of the follow-up image labeling result are achieved.

Inventors

HUANG YUEJIAO
HUANG RUINA
LIU LU
LV YULING
LI MENG
WANG PENG

Assignees

中山大学附属第八医院（深圳福田）

Dates

Publication Date: 20260505
Application Date: 20260128

Claims (8)

1. An image processing method for enhanced skin pressure injury monitoring, the method comprising: performing self-adaptive partition processing on a skin image to be processed, and establishing N local partitions, wherein the N local partitions comprise repeated partition blocks; extracting texture entropy, gradient distribution and brightness deviation characteristics of each local block, and constructing a local micro-damage response chart according to an extraction result; Constructing an adaptive enhancement function of each local block based on the local micro-injury response graph, and respectively applying different adaptive enhancement functions to carry out enhancement processing on repeated partition blocks of the same original skin image to construct a plurality of sets of enhanced images; performing interactive verification of a plurality of sets of enhanced images, calculating consistency scores of local micro-damage responses corresponding to the repeated enhanced images, performing interactive enhancement processing according to calculation results, and establishing a first enhanced image; Performing offset residual analysis of the first enhanced image, constructing a local enhanced reliability map, triggering secondary self-adaptive enhancement processing based on the local enhanced reliability map, and constructing a second enhanced image; And carrying out interactive fusion on the first enhanced image and the second enhanced image, and outputting a third enhanced image.
2. The method of image processing for enhanced skin pressure lesion monitoring according to claim 1, wherein performing an offset residual analysis of the first enhanced image, constructing a local enhanced reliability map, comprises: The method comprises the steps of calculating response offset residual errors of the same local area under different enhancement paths under the condition of spatial position alignment for enhancement results correspondingly generated by different repeated partition blocks in a first enhancement image; Calculating local texture entropy offset, gradient direction offset and brightness enhancement amplitude offset by using the response offset residual error, and combining calculation results to generate a local enhancement stability index; And carrying out credibility grading on the local area by utilizing the local enhancement stability index to form the local enhancement credibility map.
3. The method of image processing for enhanced skin pressure lesion monitoring according to claim 2, wherein the confidence rating of the local area using the local enhanced stability index further comprises: After the current partition block granularity is read, performing repartitioning treatment on the skin image according to the corresponding relation between the partition block granularity and the integral structure scale of the skin image, and establishing a macro partition, wherein the area of the macro partition is larger than that of the local partition; Based on the macro-subareas, performing intra-area continuity and response consistency verification on the enhancement results in each macro-subarea, and establishing a macro-response feature set for characterizing the enhancement stability under the macro-scale; And carrying out joint mapping analysis on the macroscopic response characteristic set and the local enhanced stability index to finish credibility grading.
4. The method of image processing for enhanced skin pressure lesion monitoring according to claim 1, wherein triggering a secondary adaptive enhancement process based on the locally enhanced reliability map, creates a second enhanced image, comprising: Carrying out credibility state division on the local area in the first enhanced image according to the local enhanced credibility map, and constructing a high-credibility enhanced area identifier, a medium-credibility enhanced area identifier and a low-credibility enhanced area identifier; Constructing constraint conditions for maintaining continuity of original skin texture structures for the low-reliability enhancement region, limiting brightness stretching amplitude and gradient enhancement intensity, and executing directional enhancement processing of stress injury associated response characteristics; Reading the enhancement result of the corresponding region in the first enhancement image in the trusted enhancement region, and performing local fine adjustment processing according to the read result; And extracting enhancement parameter distribution characteristics from the high-reliability enhancement region, establishing enhancement reference constraint, and guiding secondary self-adaptive enhancement processing of the medium-reliability enhancement region and the low-reliability enhancement region by utilizing the enhancement reference constraint to construct a second enhancement image.
5. The method of image processing for enhanced skin pressure injury monitoring of claim 1, wherein interactively fusing the first enhanced image and the second enhanced image to output a third enhanced image, comprising: Under the condition of spatial position alignment, calculating enhancement offset in local texture, gradient response and brightness response dimensions based on the first enhancement image and the second enhancement image, and constructing a local enhancement offset field representing the second enhancement image relative to the first enhancement image; performing reliability modulation processing on the local enhancement offset field based on the local enhancement reliability map to form an enhancement offset modulation field constrained by reliability; taking the first enhanced image as an enhanced main state structure, performing controlled injection processing on the enhanced offset modulation field, and introducing a correction enhancement component related to pressure damage in a second enhanced image; Based on the controlled injection processing results, a third enhanced image is reconstructed.
6. The method of image processing for enhanced skin pressure lesion monitoring according to claim 1, wherein the repeated segmentation patches are generated by respectively employing different segmentation strategies for the same skin image, the different segmentation strategies including different patch sizes, different patch start offset positions or different patch shapes, such that the same image area is covered multiple times under the different segmentation strategies to construct repeated segmentation patches having a spatial overlapping relationship.
7. The image processing method for enhanced skin pressure injury monitoring according to claim 1, wherein when constructing the local enhanced reliability map, spatial continuity constraint processing is performed on the reliability classification results of the adjacent local areas to suppress reliability mutation caused by local noise or partition boundary effect.
8. An image processing system for enhanced skin pressure lesion monitoring, characterized in that the system is adapted to implement the image processing method for enhanced skin pressure lesion monitoring according to any of claims 1 to 7, the system comprising: The image partitioning processing module is used for carrying out self-adaptive partitioning processing on the skin image to be processed, and establishing N local partitions, wherein the N local partitions comprise repeated partitioning partitions; the block feature extraction module is used for extracting texture entropy, gradient distribution and brightness deviation features of each local block and constructing a local micro-damage response chart according to an extraction result; the image enhancement processing module is used for constructing an adaptive enhancement function of each local block based on the local micro-damage response graph, carrying out enhancement processing on repeated partition blocks of the same original skin image by respectively applying different adaptive enhancement functions, and constructing a plurality of sets of enhanced images; The first enhanced image establishing module is used for performing interactive verification of a plurality of sets of enhanced images, calculating consistency scores of local micro-damage responses corresponding to the repeated enhanced images, performing interactive enhancement processing according to calculation results, and establishing a first enhanced image; The second enhanced image establishing module is used for executing offset residual analysis of the first enhanced image, constructing a local enhanced reliability map, triggering secondary self-adaptive enhancement processing based on the local enhanced reliability map, and establishing a second enhanced image; and the third enhanced image output module is used for carrying out interactive fusion on the first enhanced image and the second enhanced image and outputting the third enhanced image.

Description

Image processing method and system for enhanced skin pressure injury monitoring Technical Field The invention relates to the technical field of image processing, in particular to an image processing method and system for enhanced skin pressure injury monitoring. Background Traditional skin pressure injury monitoring relies on visual inspection and manual palpation of medical staff, and the problems of strong subjectivity, difficult identification of tiny early injury, low manual monitoring efficiency and the like exist, and skin surface images gradually become important basis for auxiliary diagnosis. However, the skin image is often interfered by factors such as uneven illumination, different shooting angles, natural change of skin texture, low contrast of early damaged areas and the like in the actual acquisition process, so that tiny or early-stage pressure damage is difficult to accurately detect. The prior image processing adopts a histogram equalization or contrast stretching enhancement strategy, although the overall visibility of the image can be improved, when local weak characteristics such as early pressure injury and the like are processed, global enhancement can amplify irrelevant noise or normal skin texture, interfere identification of a real injury area, the influence of pressure degree, subcutaneous structure and illumination condition on the skin of different parts has high heterogeneity, a single enhancement function is difficult to adapt to the characteristics of each area, the difference between the early injury area and surrounding healthy tissues is very fine, the conventional enhancement easily causes characteristic inundation or distortion, and the local enhancement based on area segmentation mostly adopts non-overlapping partitions, is easy to generate enhancement discontinuity at the block boundary, can not completely cover the tiny injury, and is difficult to ensure the stability and reliability of enhancement under complex conditions. Therefore, in the related art at the present stage, the conventional image enhancement cannot adapt to the local heterogeneity of the skin image, and the early weak injury features are difficult to extract effectively and the reliability is low. Disclosure of Invention By providing the image processing method and the system for enhanced skin pressure injury monitoring, the technical problems that the conventional image enhancement cannot adapt to the local heterogeneity of skin images, the early weak injury features are difficult to extract effectively and the reliability is low in the prior art are solved, and the technical effects of realizing the accurate enhancement processing of the local micro injury features so as to ensure the reliability of the follow-up image labeling result are achieved. The application provides an image processing method for enhanced skin pressure injury monitoring, which comprises the steps of carrying out self-adaptive partitioning processing on a skin image to be processed, establishing N local partitioning blocks, extracting texture entropy, gradient distribution and brightness deviation characteristics of each local partitioning block, constructing a local micro-injury response graph according to an extraction result, constructing self-adaptive enhancement functions of each local partitioning block based on the local micro-injury response graph, carrying out enhancement processing on the repeated partitioning blocks of the same original skin image by respectively applying different self-adaptive enhancement functions, constructing a plurality of sets of enhancement images, carrying out interactive verification on the plurality of sets of enhancement images, calculating consistency scores of local micro-injury responses corresponding to the repeated enhancement images, carrying out interactive enhancement processing according to a calculation result, establishing a first enhancement image, carrying out offset residual analysis on the first enhancement image, constructing a local enhancement reliability graph, triggering secondary self-adaptive enhancement processing based on the local enhancement reliability graph, establishing a second enhancement image, carrying out interactive enhancement processing on the first enhancement image and the second enhancement image, and outputting a third enhancement image. In a possible implementation manner, offset residual analysis of a first enhanced image is executed to construct a local enhanced reliability map, and the method comprises the steps of calculating response offset residuals of the same local area under different enhanced paths under the condition of spatial position alignment on enhancement results generated by different repeated partition blocks in the first enhanced image, respectively calculating local texture entropy offset, gradient direction offset and brightness enhancement amplitude offset by using the response offset residuals, combining calculation result