Color-Based Skin Detection and its Application in Video Annotation
Christian Liensberger, Martin Kampel · 2009
Skin detection in visual data cannot be solved by analyzing the low level image features only: In an extensive online experiment we are able to show that even humans are not able to detect skin color reliably without knowing the context of their perception. To compensate for this ma- jor drawback of many approaches, we combine a state of the art recognition algorithm with color model based skin detection. Detected faces in videos are the basis for adap- tive skin color models, which are propagated throughout the video, providing a more precise and accurate model in its recognition performance than pure color based approaches. The approach is able to run in real-time and does not need prior data-specific training. We received challenging online videos from an online service provider and use additional videos from public web platforms covering a grand variety of different skin colors, illumination circumstances, image quality and difficulty levels. In an extensive evaluation we estimated the best performing parameters and decided on the best model propagation techniques. We show that adap- tive model propagation outperforms static low level detec- tion.