Image Preprocessing Using Reference Similarity for Biometric Color Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image preprocessing methods struggle to accurately determine whether an image is suitable for biometric color models, often leading to incorrect processing and recognition results due to the use of pseudo-color images that resemble grayscale images.
Innovation Solution
A method and system that calculate a reference similarity between channels of an image to determine its suitability for a biometric color model, using threshold values to differentiate between color and grayscale images, and employing a classifier to further assess images in a fuzzy zone, thereby preventing the misuse of biometric color models on unsuitable images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image preprocessing is performed using biometric color model, then person recognition accuracy is improved, but false positives occur when pseudo-color images resembling grayscale images are incorrectly processed
Solution Approach 1:
The patent applies preliminary action by performing image suitability assessment before the actual person recognition processing. The system calculates reference similarity between color channels and compares it with threshold values to determine whether the image is suitable for biometric color model processing, thereby preventing incorrect processing of pseudo-color images beforehand
Solution Approach 2:
The patent introduces an intermediary mechanism - the reference similarity calculation and threshold comparison system - that acts as a mediator between the image input and the biometric color model processing. This intermediary assesses image suitability and directs appropriate processing paths, preventing pseudo-color images from entering the biometric color model processing pipeline
2Measurement precision
If reference similarity calculation is performed to determine image suitability, then processing accuracy is improved, but processing time increases due to additional calculation steps
Solution Approach 1:
The patent applies partial action by implementing a two-stage assessment process. First, a quick reference similarity calculation with threshold comparison is performed for all images. Only images falling into the fuzzy zone (where similarity is close to threshold) undergo further classification. This partial application of excessive action ensures high accuracy while minimizing time loss on clearly suitable or unsuitable images
Data Source
AI summary
Preprocessing images is disclosed including determining a reference similarity of an image to be processed, and in the event that the reference similarity is less than or equal to a first preset threshold value, processing, based on a biometric color model, the image to determine a biometric color area of the image.