基于独立分量分析的运动目标精确检测

运动目标检测与跟踪是计算机视觉的难点问题.针对传统方法存在抗噪声、抗抖动性能弱等问题,将独立分量分析(ICA)的算法应用到目标检测与前景图像去噪环节中,目的是提高运动目标检测的鲁棒性和精确性,为后续处理提供尽可能多的目标信息.实验表明,ICA用于目标检测具有较高的鲁棒性,同时ICA用于前景图像去噪比传统的去噪方法具有更好的效果,较好地保留了目标的特征信息.

Accurate Moving Objects Detection Based on Independent Component Analysis

The techniques of object detection and tracking are the focal points in the fileds of computer vision.There are some drawbacks in the traditional methods of moving object detection and tracking,such as easily disturbed by noise and trembling.Aming at these problems,this thesis proposes a method which uses the algorithms of Independent Component Analysis（ICA） in the segment of target extraction and image denoising.The goal is to improve the robustness of detecting and the accuracy of denoising so that we can provide more information of the target for the follow-up processing.The experiment results show that it is a robust and accuracy method,and it preserves more feature information of the target than other traditional methords.