Aiming at the high false alarm rate of current local contrast methods for target detection in complex backgrounds,a LOG filtering based bidirectional weighted local contrast infrared small target detection algorithm is proposed.Firstly,after smooth-ing and fine-tuning the infrared images under different backgrounds by changing the standard deviation s value of Gaussian filter-ing,the Laplacian operator is used to perform convolution operation to quickly extract effective target pixels under complex back-grounds.Then,considering that the target pixels have different feature information in the orthogonal and diagonal directions,the bi-directional local contrast is constructed by using the product of the local contrast in the orthogonal direction and the diagonal direc-tion and weighted to further improve the salience of the target.Finally,the real target to be detected is obtained through adaptive threshold segmentation.Experimental results show that the proposed algorithm can detect objects of different sizes,not only avoid the block effect caused by multi-scale algorithms,but also improve the detection rate of objects in complex backgrounds.
关键词
高斯-拉普拉斯滤波/红外小目标/双向加权局部对比度/目标检测
Key words
Gaussian-Laplacian filtering/small infrared target/bidirectional weighted local contrast/target detection
维普
万方数据