用于行人检测的处理库

Hog-Processing Histogram of Oriented Gradients (HOG) is a popular method for object detection, particularly suitable for pedestrians detection. The method is based on the evaluation of histograms calculated on the basis of the orientation of the gradients of the input image. The idea is that the edges and the shapes of objects can be well characterized by the intensity of individual local gradients. Computation of each histogram is obtained by dividing the image into a grid of cells and for each of them is computed a histogram of gradients. Then the cells are grouped into regions called blocks. In addition, to release the response by the terms of brightness, can be useful normalize individual blocks. From information of the individual blocks you get a descriptor that will be used for detection. The above operations are performed on windows of finite size that analyze the image at multiple scales. For each window is then obtained a descriptor that is then