Algorithm For Weak And Small Target Detection Under Complicated Background

This paper makes a comprehensive and in-depth study on infrared weak and small targets detection under the complicated background of sky.Because of the characteristic difference between weak and small target, large area background and random noise in the infrared image, the design of target detection concentrates on the background suppression. Based on summarizing several classical algorithms home and abroad, two pretreatment methods are brought forward in this paper to improve performance in suppressing the background. One is detection method based on singular value decomposition (SVD) which selects effective numbers of singular value by adopting statistical criteria, and reconstructs a new image matrix to achieve background estimation. The other is modified bilateral filter (MBF) background suppression detection. For modified bilateral filter detection method, not only abstract the advantage of traditional bilateral filter that spatial closeness and photometric similarity between pixels are taken into consideration, but also add the local constraint condition of gradient standard deviation to enhance the effect in background suppression.In order to solve the difficulty in separating random noise from weak and small target which have the similar feature in the single frame, this paper makes use of the continuous property of motion position of targets and stability of pixels' gray-scale value in the sequence images, and proposes the target-motion detection with image sequences. Ultimately achieve reliable infrared weak and small targets detection under the condition of complicated background of sky and SNR no less than 1.6.