Picture Type Identification Using Histogram Ratios
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying picture types on mobile terminals, such as distinguishing between composite and nature pictures, are inefficient due to high computational requirements, making them unsuitable for mobile devices.
Innovation Solution
The method involves acquiring histograms of respective color channels in a preset color space, calculating ratios between adjacent gray-scale pixels, and determining the picture type based on the number of decimal places in these ratios, with optional noise filtering and feature library matching to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If entropy calculation method is used to identify picture type, then identification accuracy is improved, but computational complexity increases enormously
Solution Approach 1:
The patent extracts only the most significant feature (brightness histogram) from the complete picture data, ignoring color information and other attributes. By taking out only the essential brightness distribution characteristics, the method achieves accurate identification without the enormous computational complexity of analyzing all picture data through entropy calculation.
Solution Approach 2:
The patent segments the picture analysis into two independent parts: brightness channel extraction and histogram calculation. This segmentation separates the complex task of full picture analysis into simpler, more manageable steps, reducing computational load while maintaining identification accuracy through focused analysis of brightness characteristics alone.
2Measurement precision
If comprehensive picture analysis is performed, then identification accuracy is improved, but processing time increases
Solution Approach 1:
The patent extracts only the brightness histogram from the complete picture data, disregarding color channels and other attributes. This selective extraction maintains sufficient identification accuracy while dramatically reducing processing time compared to analyzing all picture characteristics comprehensively.
Solution Approach 2:
The patent applies partial action by analyzing only the brightness channel rather than all picture data. This partial analysis is sufficient for accurate picture type identification, avoiding the time-consuming complete analysis while maintaining adequate accuracy through focused examination of brightness distribution characteristics.
Data Source
Figure 1~2b
Figure 2c~3c
Figure 4~5
AI summary
The present invention relates to a method and apparatus for identifying picture type, wherein the method for identifying picture type includes: acquiring (S101) histograms of respective channels in a preset color space of a picture; calculating (S102) ratios between numbers of adjacent gray-scale pixels in respective channels according to the histograms; when the ratios satisfy a preset condition, determining (S103) that the picture is a nature picture; and when the ratios do not satisfy the preset condition, determining (S104) that the picture is a composite picture. The embodiments of the present invention need no large number of computations during the entire identifying procedure, thus may be suitable for identifying picture type in a mobile terminal such as a mobile phone, and has a strong applicability.