Infrared low, slow and small target detection based on neighborhood characteristics

The infrared search system has the advantages of not emitting electromagnetic wave, strong anti-electromagnetic interference ability and high precision of target indication, and has a good application prospect in the field of low, slow and small target detection. At present at home and abroad, the target detection algorithm of infrared search system usually extracts the suspected target by using the current image and the background image registration and difference, which is often necessary to save the panoramic background images in the large storage space, and the engineering application of high precision real-time image registration algorithm is also difficult. For the above questions, a infrared low, slow and small target detection method was designed for the infrared search system. Through the processes of morphological filtering, extracting suspected targets with edge detection method, eliminating background interference with target neighborhood eigenvalue statistical method and correlating target information based on multi-frame image, it can effectively eliminate the interference of ground objects, clouds and birds, while accurately detecting the UAV's target in the image. Experimental results show that this method has higher target detection probability and lower false alarm rate than the traditional LCM algorithm. Moreover, it does not involve image difference, and has the advantages of low demand for hardware resources and good real-time performance, etc., and has high engineering application value.