Automated Front View Background Separation Using Color Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for separating a front view and background in image recognition techniques are not fully automated and are prone to errors due to background noise, edges, or patterns, requiring user-defined regions and resulting in lower recognition rates.
Innovation Solution
A method that divides image pixels into groups based on color similarity using the CIE Lab color space, estimates the front view position by detecting edges and calculating circular scores, and separates the front view from the background without user interaction, utilizing techniques like the canny edge detector and chi-square histogram comparison.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If user-defined regions are used for front view and background separation, then the separation can be controlled, but the automation level decreases and requires user interaction
Solution Approach 1:
The system performs front view and background separation automatically without requiring user input or definition of regions. The algorithm analyzes image characteristics autonomously to identify and separate the front view object from the background, eliminating the need for user interaction while maintaining separation accuracy.
2Reliability
If background noise, edges, or patterns are present in the image, then the image content is preserved, but the recognition rate decreases due to misjudgment
Solution Approach 1:
The image is divided into multiple regions based on color similarity and edge detection. By segmenting the image into distinct pixel groups and regions, the algorithm can identify the front view object separately from the background, even when the background contains noise, edges, or patterns that would otherwise interfere with recognition.
Solution Approach 2:
The patent introduces intermediate processing steps including color space conversion to CIE Lab, edge detection, and circular score calculation. These intermediary processes act as filters that eliminate the harmful effects of background noise and patterns before the final separation decision is made, thereby improving recognition reliability.
3Measurement precision
If boundary-based recognition algorithms are used, then object boundaries can be detected, but the recognition rate is lowered due to misjudgment of object boundaries
Solution Approach 1:
The patent changes the parameter space from traditional boundary-based detection to color similarity-based segmentation in CIE Lab color space. By transforming the detection criterion from edge sharpness to color homogeneity within regions, the algorithm achieves more accurate object boundary detection that is not susceptible to misjudgment by conventional boundary algorithms.
Data Source
AI summary
A method of initially estimating a front view portion of a photographed image and separating the photographed image into a front view and a background without user interaction and apparatus performing the method are provided. The method of separating a front view and a background of an image includes dividing one or more pixels included in a photographed image into pixel groups according to color similarity between the pixels, estimating the position of the front view in the image divided into the pixel groups, and separating the front view and the background based on the estimated position of the front view. The method automatically separates the front view and the background of the image without a user input.