Skin Tone Detection Using Adaptive Histograms for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current skin tone detection techniques face limitations such as limited accuracy, high computational demands, and inability to perform real-time processing, making them inadequate for widespread adoption in applications like video coding and human-computer interaction.
Innovation Solution
The proposed solution involves using a skin tone detection system that employs multiple static and dynamic skin probability histograms in the Yrg and YUV color spaces, combined with adaptive luma histograms and motion cues, to accurately identify and track skin tones in video frames, enabling efficient coding and improved detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing skin tone detection techniques are used, then detection capability is provided, but accuracy is limited and computational demands are high
Solution Approach 1:
The patent segments skin tone detection into multiple stages: initial skin probability map generation using color space histograms, followed by region-of-interest identification, and finally refined detection within those regions. This segmentation allows accurate detection while reducing overall computational complexity by focusing resources on relevant areas.
Solution Approach 2:
The patent employs dynamic skin probability histograms that adapt to varying lighting conditions and skin tones. The system dynamically adjusts detection parameters based on scene analysis, enabling high accuracy across diverse conditions without requiring overly complex fixed algorithms.
2Productivity
If existing skin tone detection techniques are used, then detection capability is provided, but real-time processing is not achievable
Solution Approach 1:
By segmenting the detection process into rapid initial filtering using color histograms and subsequent refined detection only in identified skin regions, the patent achieves real-time processing speeds while preserving detection accuracy. The majority of frames are processed quickly through the initial stage.
Solution Approach 2:
The patent applies full detection accuracy only to regions identified as potential skin areas, rather than processing entire frames at maximum detail. This partial action approach maintains accuracy where needed while achieving real-time performance through reduced overall computation.
3Measurement precision
If comprehensive skin tone detection is performed, then detection accuracy is improved, but computational demands increase
Solution Approach 1:
The patent extracts and processes only the most relevant features for skin tone detection—specifically color space histograms and luminance information—while discarding less relevant data. This extraction approach maintains detection accuracy while significantly reducing computational energy requirements.
Solution Approach 2:
The system applies comprehensive detection analysis only to identified skin regions rather than entire video frames. This partial application of full detection capability preserves accuracy for skin tones while reducing overall energy consumption proportional to the fraction of skin pixels in typical scenes.
Data Source
AI summary
Techniques related to improved video coding based on skin tone detection are discussed. Such techniques may include selecting from static skin probability histograms and/or a dynamic skin probability histogram based on a received video frame, generating a skin tone region based on the selected skin probability histogram and a face region of the video frame, and encoding the video frame based on the skin tone region to generate a coded bitstream.