摘要
采用光流算法对运动目标进行识别跟踪,其优点是能够适应复杂的背景条件,并且能保证目标分割的完整性,但现有的按照光流矢量对目标进行跟踪的算法有明显的局限性:运算量过大,并且不适用与运动特征复杂的目标。对现有算法进行改进,采用均值平滑算法和基于光流绝对值的区域分割算法,可以有效解决这两个问题。
Using optical flow algorithm for identification and tracking moving targets, the advantage is the ability to adapt to the com- plex background conditions, and can ensure the integrity of the target partition, but the existing target tracking algorithm based on op- tical flow vector has obvious limitations： excessive operation, and does not apply and movement characteristics of complex targets. Im- provements to existing algorithms, using the pyramid optical flow-based smoothing algorithm and the absolute value of the region seg- mentation algorithm can effectively solve these two problems.
出处
《微计算机信息》
2012年第10期421-423,共3页
Control & Automation
关键词
光流
运动目标
图像分割
Key word： Optical flow
Kinetic target
Image segmentation
