Image Identification Using Texture and Color Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image search technologies rely on text-based metadata, limiting the ability to directly search for similar images based on their content, as they do not effectively utilize texture and color attributes for content-based image retrieval.
Innovation Solution
An image identification method that extracts texture and color features from images, generating parameters through discrete cosine transformation and chromaticity analysis, allowing for direct comparison and similarity determination between images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If text-based metadata search is used for image retrieval, then the search system is simple to implement, but the ability to directly search for similar images based on content is limited
Solution Approach 1:
The patent transforms image content into quantifiable parameters by extracting texture parameters (through discrete cosine transformation of gray-level images) and color parameters (through chromaticity analysis). This parameter transformation enables direct content-based comparison and similarity determination, resolving the limitation of text-based metadata search while maintaining systematic implementation.
2Measurement precision
If texture and color feature extraction is performed on images, then content-based image retrieval accuracy is improved, but the processing complexity increases
Solution Approach 1:
The patent divides the image analysis process into distinct segments: gray-level image generation, discrete cosine transformation for texture parameters, and chromaticity analysis for color parameters. This segmentation allows each feature extraction task to be handled independently with dedicated algorithms, managing complexity while achieving comprehensive content analysis.
Solution Approach 2:
The patent introduces intermediate representations to simplify processing: gray-level images serve as intermediaries between color images and texture parameters, while chromaticity values act as intermediaries between RGB values and color parameters. These intermediaries decouple complex transformations into manageable steps.
3Measurement precision
If discrete cosine transformation and chromaticity analysis are used for feature extraction, then image identification accuracy is improved, but the computational time increases
Solution Approach 1:
The patent extracts only the essential and discriminative features from images: texture parameters from discrete cosine transformation coefficients and color parameters from chromaticity analysis. By extracting only relevant features rather than processing entire images, the system achieves high identification accuracy with reduced computational burden compared to full-image comparison.
Data Source
AI summary
An image identification method, an electronic device with image identification function and a computer program product executing the image identification method with a software program are provided. The image identification method comprises steps of: proceeding texture feature extraction on a color source image to obtain a plurality of texture parameters; proceeding color feature extraction on a color source image to obtain a plurality of color momentums; and weighting the plurality of texture parameters and the plurality of color momentums to obtain an image identification parameter corresponding to the color source image.