Abstract
The purpose of this paper is the automatic detection of human beings in a static image. We propose the method of human detection by using the feature map based on the visual attention model. The proposed algorithm uses the features of orientation and gradient magnitude for human detection. We also propose to use the symmetric filter to incorporate bilateral symmetry of human. The use of Gaussian pyramids for making the feature map enables the detection of the most salient feature location. Finally, the human detection is performed by using the human feature map, that is obtained by integrating each feature, and nonlinear support vector machines.
