Image Processing Apparatus Fuzzy Region Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-class segmentation methods, such as Recursive Neural Networks, often misclassify regions with non-obvious classification scores, leading to incorrect classification results due to close scores between classes, particularly affecting fuzzy regions which are not self-adaptive.
Innovation Solution
An image processing method that identifies fuzzy regions by extracting them based on confidence thresholds or Graph-cut algorithms and updates their classification confidence using neighborhood information, allowing for re-classification to improve precision without affecting obvious regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the Recursive Neural Network method is used to classify regions based on extracted features and trained models, then the classification process is automated and efficient, but the classification precision deteriorates for fuzzy regions where the highest score is not obviously higher than the second highest score
Solution Approach 1:
The patent segments the image regions into two categories: obvious regions and fuzzy regions. This segmentation allows the system to apply different processing strategies to different region types, maintaining efficient automated classification for obvious regions while applying enhanced analysis to fuzzy regions to improve their classification precision.
Solution Approach 2:
The patent applies local quality by treating obvious regions and fuzzy regions differently. For obvious regions, the standard RNN classification is used to maintain efficiency. For fuzzy regions, additional processing steps including neighborhood information integration and iterative re-classification are applied to improve precision locally without affecting the overall processing efficiency.
2Speed
If the classification is performed solely based on the highest confidence score without considering neighborhood information, then the processing speed is maintained, but the classification accuracy deteriorates for regions with close scores between classes
Solution Approach 1:
The patent performs preliminary identification of fuzzy regions by comparing confidence scores before applying enhanced processing. This preliminary action allows the system to maintain fast processing for obvious regions while preparing fuzzy regions for iterative re-classification, balancing speed and accuracy.
Solution Approach 2:
The patent implements feedback mechanisms through iterative re-classification of fuzzy regions. The system uses neighborhood information to adjust and refine classification results, allowing the classification accuracy to improve while maintaining processing speed through targeted iteration only on fuzzy regions rather than all regions.
Data Source
AI summary
The present invention discloses an image processing evaluating apparatus and image processing method. The processing apparatus comprising: a confidence generation means for generating a classification confidence for each region in the image, the classification confidence represents the probability of an region belonging to a predefined class; and a classification means for classifying the regions in the image, which are obvious to be classified by their classification confidences, to respective classes based on the calculated confidences. The image processing apparatus further comprising: a fuzzy region extraction means for extracting one or more regions, which are not obvious to be classified by their classification confidences, as fuzzy regions; and a confidence update means for updating the classification confidence for each fuzzy region based on the classification confidences of adjacent regions thereof, wherein the classification means further classifies the fuzzy regions to respective classes based on the updated classification confidences.