Knapsack Model for Pixel Based Skin Detection

Skin detection has different applications in computer vision such as face detection, human tracking and adult content filtering. One of the major approaches in skin detection is pixel based skin detection which mark pixels as skin pixels based on their color. In this paper, we formally define the pixel based skin detection problem and model skin detection as a Knapsack problem. Then we formulate the maximum accuracy which can be reached by training on a given data set. Finally, we study one of the common pixel based skin detection methods in our Knapsack model.