Object detection method and system through constructing feature pool with feature templates

The invention discloses an object detection method and system through constructing a feature pool with feature templates. The method includes the steps of processing an original image to obtain a grayscale image, a gradient amplitude image and a gradient histogram, setting one or more feature templates of different meshes, moving the feature templates on the grayscale image, the gradient magnitude image and the gradient histogram respectively to obtain a feature pool of image features, training the image features in the feature pool using an adaboost algorithm to obtain an optimal object detection feature, and finally performing object detection according to the optimal object detection feature. According to the invention, more features of the image can be obtained through image feature extraction and feature pool construction for the images of different nature by using the different feature templates, so as to fully exploit useful information of the image and distinguish the features of different object targ