Abstract: Based on the features of dim infrared point targets, this paper adopts the method of threshold comparison before high pass filtering and the threshold segmentation after filtering, in order to separate the infrared point targets efficiently from low s/n and complex background. At the same time, spot targets are detected by combining point targets that have been separated. This algorithm can not only reduce the amount of operation but is also applicable to some complex cases where the point targets and spot targets exist at the same time.