Laser Radar, Infrared Point Target Image Preprocessing Technology Research

The research on point target detection technology has an important military significance and value to enhance the survivability of regional defense system and the effective retaliatory capability and the control of the air in a future war.The image of point target in the imaging area is only several or dozens of pixels. And its strength is relatively weak and drowned in the noise, no obvious size, shape, texture, and other information. The infrared remote imaging objectives are often hidden in a highly structured clutter in the background, the target signal amplitude is relatively weak to the background noise and clutter, which has low signal to noise ratio. Therefore, point target detection of remote imaging is hot and difficult in infrared image detection and processing in recent years.In general, point target detection of Infrared image can be divided into three phases, followed by image preprocessing, target segmentation and target detection. This paper mainly focuses on image preprocessing.The main purpose of image preprocessing is to improve the quality and SNR of the image which is transmitted to processor, suppress background and clutter as much as possible. The results of their treatment have a direct impact on system detection performance and real-time of entire algorithm.Based on a variety of characteristics of the commonly used algorithms, denoising theoretical analysis, and carried out the research work of simulation, and several preprocessing algorithm which can be used to point target detection algorithm are proposed. At the same time, the infrared image that is provided is processed:(1).The denoising technology based on median filtering + mathematical morphology.(2).The denoising technology based on the wavelet high-frequency coefficientsreconstruction of BayesShrink threshold.(3).The denoising technology based on wavelet high-frequency coefficientsreconstruction of the threshold formulaσ(2lgM×N)^1/2/lg(j+1).(4) The high-pass filtering based on the Median difference. The point target which has fewer pixels can be effectively highlighted. (5) The high-pass filtering based on the Wiener difference.In this paper, the relevant research and analysis of simulation results to the above denoising technology show that: the denoising SNR of using the first three algorithms can be reached requirements, which have the better denoising effect and certain extent practical value. The SNR can not be reached to the requirements after median difference filtering and wiener difference filtering, but the difference thinking can be used during the wavelet denoising for better result in future work.