Computer vision based pedestrian detection has become one of the hottest topics in the domain of intelligent vehicle because of its potential applications in driver assistance systems. In this paper, we put forward an improved real-time detection and Localization method for pedestrian. We select Haar-like features, sums up nine categories rectangle features which highlight the different characteristics of pedestrian appearance, and use integral image to speed up the calculation for rectangular features. Meanwhile, we introduce an optimization method of updating weight into the traditional AdaBoost algorithm, design a cascade classifier, and obtain a better pedestrian classifier with accurate recognition by faster training, which realize the real-time regional rapid detection of pedestrians. Experimental results show that the improved training process has short training time, high accuracy, and strong real-time performance.