The method based on the integral image and the haar-like feature for detecting body limbs is proposed for video images. Haar-like feature is suitable for detecting rectangular or rectangular-like image areas because of its inherent characteristics, and the integral image can quickly locate the body by the haar-like feature in the image. The integral image is the original image for video. After edge detection template is convoluted with each frame of the original video, they are added to obtain the integral image. The body physical location can be calculated by the integral image and using haar-like feature according to the set threshold. The experiment results show the algorithm is better than the background subtraction.%为了实现视频序列图像中运动人体肢体的检测,提出了一种基于积分图像和类哈尔特征的检测方法,类哈尔特征(Haar-like feature)因其固有的特点,适用于检测矩形或类矩形的图像区域,积分图像的特点是可以利用类哈尔特征快速定位人体肢体在图像中的位置.积分图像用于表示视频中的原始图像,边缘检测模板与原始视频每一帧进行卷积之后,通过累加计算可以得到该图像的积分图像,然后利用类哈尔算子,根据设定的阈值,即可以准确定位人体肢体位置.实验结果表明:该方法较背景减除法检测人体肢体准确.
展开▼
