Monitoring Region Recognition Using Prototype-Guided Image Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods suffer from low recognition accuracy when clarity of images is low, leading to inaccurate feature extraction and recognition of target parts.
Innovation Solution
Perform semantic segmentation on images to obtain a first region feature, determine a second region feature based on a dependency relationship with pre-constructed prototypes, and recognize feature information using both features to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only feature extraction is performed on images with low clarity, then the processing speed is maintained, but the recognition accuracy deteriorates
Solution Approach 1:
The patent segments the image processing into multiple stages: obtaining initial features from the original image, performing semantic segmentation to get first region features, and then fusing these with the initial features to obtain second region features. This segmented approach allows each stage to contribute specifically to improving recognition accuracy while maintaining manageable processing complexity through modular operations.
Solution Approach 2:
The patent performs preliminary semantic segmentation to obtain first region features before final feature fusion. This preliminary action of segmenting the image into meaningful regions with attribute information prepares the data structure in advance, making the subsequent feature fusion more effective and improving overall recognition accuracy without significantly increasing final processing complexity.
2Measurement precision
If semantic segmentation and prototype-based feature determination are performed, then recognition accuracy is improved, but processing time increases
Solution Approach 1:
The patent applies local quality by performing semantic segmentation specifically on the monitoring region of the target part rather than the entire image. This localized approach extracts first region features only where needed, and then fuses these with initial features to obtain second region features. This reduces the overall processing time while maintaining high feature extraction accuracy in the critical monitoring region.
3Reliability
If multiple features (first region feature and second region feature) are fused for recognition, then recognition accuracy is enhanced, but the complexity of the recognition system increases
Solution Approach 1:
The patent merges the first region features (from semantic segmentation) with the initial features (from the original image) to obtain second region features. This combining of complementary feature sources enhances recognition reliability by integrating multiple information streams. The merging is achieved through feature fusion operations that consolidate the features into a unified representation, managing system complexity through systematic integration rather than separate independent systems.
Data Source
AI summary
An image processing method comprises: acquiring a plurality of images, wherein display content of the images comprises at least a monitoring region of a target part of an object to be monitored; performing semantic segmentation on the images to obtain a first region feature of the monitoring region in the images; determining a second region feature of the monitoring region based on a dependency relationship between a plurality of pre-constructed prototypes and the first region feature, wherein different prototypes are used to represent different types of monitoring regions; and identifying feature information of the monitoring region based on the first region feature and the second region feature to determine an identification result of the monitoring region. The present application solves the technical problem of low identification accuracy when identifying images to be monitored in related art.