The Research Of Small Moving Detection In Complex Background

In the infrared small target images, because the pixels are very lack and short of structure information, the research about it is always a knotty problem.In chapter two, after analyzed and compared many background restrain algorithms uses a method of background restrain which combined wavelet transform with morphology filtering. The method obviously improved the efficiency of small target detection.This paper presents a novel method of small moving target detection. The method combines the information in frames with optical flow. It is divided into preprocessing technology, feature detection and Lucas-Kanade optical flow based on pyramid construction. The algorithm works well for small moving target detection as demonstrated by experimental results on real image sequences.This paper presents an efficient small target detection algorithm for infrared sequence image based on particle filter. The algorithm mainly uses the statistic values of the target's features image to change the weight of the particles which are produced by Monte Carlo sampling, and the target is detected and tracked based on the control of these weights. The difference image is used as the feature image. The experimental results also demonstrate that particle filter algorithm suitable for the non-linear movement target detection regardless of the disturbance of serious non-Gaussian noise.