Infrared Weak Small Target Image Enhancement And Target Match

The enhancement of infrared images containing dim targets is of great importance for automatic target recognition and infrared target tracking. Unfortunately, there are many difficulties in such images, including low signal-to-noise ratio (SNR), low contrast and high similarity between target and noise. All of them make it challenging for further processing of images, like target recognition and extraction, tracking and detection etc. Therefore, the research along this direction is essential to infrared image processing, and has been an active topic in the area.This work can be roughly divided into two parts, i.e., the enhancement of infrared images containing dim targets, and the searching and matching for an infrared area target.As the first part, an enhancement approach to improve the SNR and contrast as well as to remove the noise is presented. Before the approach is detailed, some basic concepts of infrared images and evaluation criteria for image enhancement are briefly introduced, e.g., the characteristics of both the target and background. Then, several types of noise signals involved in infrared images are described. In addition, this work gives the advantages and limitations of some general filters for noise removal, such as histogram-based method, morphological filter, and Retinex algorithm. By taking into account the characteristics of infrared images, a new method based on Retinex algorithm is presented. It can highlight the target while suppressing the background. Experimental results demonstrate this method is effective in enhancing the infrared images containing dim targets. Furthermore, this method is implemented on DSP.Next, regarding the issues of searching and matching for an area target,the paper also studies a highly efficient searching and matching algorithm. Based on the study of quantum-behaved particle swarm optimization (QPSO) algorithm, the paper provides a two dimensional image matching algorithm using Hausdorff distance and quantum-behaved particle swarm optimization (QPSO) algorithm. It is proved that this algorithm is better than the genetic algorithm in both efficiency and robustness. According to the simulation experiments, the proposed method shows good performance in area target searching and matching in infrared images.