foreground objects are segmented by using a per-pixel color-based hybrid gaussian model as the background updating method, and tracked by template matching.
以每点色彩信息的混合高斯模型来实现对视频图像的背景估计,从而实现对运动目标的检测。
on this basis, combined with one sort of background updating model, vehicle tracking algorithm is presented through using two-dimension motion estimation technique in dynamic image processing.
在此基础上,使用动态图像处理技术中的二维运动估计技术,结合背景更新模型提出了一种车辆跟蹤算法。
the proposed new background updating method can update the background exactly and quickly, and the detection result of moving objects is good without background image.
本文提出的算法鲁棒性好,在没有背景先验知识的情况下对运动物体的检测取得了很好的效果。
in feature extraction, a difference motion detection algorithm of adaptive background updating is proposed, which has a good robustness to the changing of background and illumination.
在特征提取方面,提出了一种自适应背景更新的差分运动检测算法,该算法在运动对象轨迹的检测中对背景的抖动和光照条件的变化具有较好的鲁棒性。
main work in low level vision include researching methods of vehicle detecting and segmentation, as well as techniques of image pretreatment, background updating and shadow removal.
低层视觉中,主要研究了车辆的检测与分割的方法,还包括图像的预处理、背景更新以及阴影消除等技术的研究;