Research Of Small Dim Target Detection Method In The Infrared Image

The infrared imaging technology is widely used many situations, such as the search and rescue of people、the analysis of pathology in medical field and the detection for hidden danger of fire, of which all need the accurate information of targets. And the detection of small dim target in infrared images is a key technology of getting accurate information of targets in the infrared imaging. So the researching of a suitable method for the detection of small dim target in infrared images can not only offer some important reference for every area which are based on infrared imaging technology, but can also contribute to the study on infrared imaging technology.Firstly, analysis the background、the meaning and the purpose of the detection of small dim target in infrared images one by order in the chapter one. In order to understand and comprehend the situation of the detection of small dim target in infrared images at present, introduced the current situation of the detection of small dim target in infrared images in domestic and overseas. Then studied the way of getting infrared images and the process of imaging by infrared equipments, and then studied the feature of infrared images which contain small and dim targets, and analysis the process of the detection of small dim target in infrared images. Based on the study of the background technology and basic theory above, the method of pretreatment for the detection of small dim target in infrared images was studied in the chapter three, in this chapter a wavelet threshold algorithm based on vector quantization model was put forward, after the contrast of other six basic methods. In the chapter four, first making use of the image segmentation method, transformed the gray image to binary image, then analysis the feasibility of mathematical morphology method and pipe filter method, and studied the implementation method of these methods, finally putted forward a method of small dim target detection in infrared images, which make used of neural network improve the mathematical morphology method. And this method can be used in targets preliminary detection step and targets identify in the end. In this new method the error rate of the targets preliminary detection was improved by the neural network, what’s more, the traditional pipe filter method was improved based on the center point of a target, and the accuracy of targets identify in the end was increased too. Through comparison experimental simulation with traditional method, the advantages of infrared dim small targets detection method based on neural network, which is put forward in this paper, was proved.