Automatic Recognition of Weak Point Targets in Infrared Images Based on Recursive Tagging
To accurately identify weak point targets in infrared images,an automatic recognition method of weak point targets in infrared images is proposed based on recursive marking algorithm.Smoothing filter and region growing segmentation method are used to com-plete the preprocessing of the infrared image to be recognized.The binary image marking mode based on recursion is applied to scan all infrared image pixels and define corresponding marking values to form multiple image connected regions.Geometric features and cen-troid features are extracted for each connected domain,and applied to sparse representation classifier and convolutional neural network classifier to automatically generate weak point target recognition results.The experimental results show that the mAP values of sea sur-face infrared image,cloud layer infrared image and pure sky infrared image are 0.95,0.96 and 0.91 respectively,which shows that the proposed method has a good automatic recognition effect for weak point targets in infrared image.
NETL
NSTL
万方数据
