A part-based rotational invariant hand detection
Jisu Kim, Jeonghyun Baek, Euntai Kim · 2013
This paper proposes part-based rotational invariant hand detection in complicated environment. The proposed method consists with 4 steps: head detection, back projection, hand rotation and hand detection. To improve the hand detection performance, the human head information is used for generating effective hand ROIs. Generated hand ROIs are verified using Histogram of Oriented (HOG) feature and Support Vector Machine (SVM) classifier. In the experiment, the proposed method is compared with previous methods and detection performance and computational time are evaluated.