Infrared Target Detection Based On Combination Of Region And Edge

With the development of infrared imaging technology, target detection and recognitionbased on infrared image have been more widely used. Different from visible light imaging，the infrared images obtained with low contrast and blurred edges, etc， which causesdifficulties for target segmentation, detection and identification. So studying a more effectiveinfrared image segmentation technique is an important issue in image processing.This paper first introduces the characteristic of the infrared image. Then we propose aimage segmentation algorithm based on combination of region and edge solving the smalltarget and weak edge problem.CV model is a commonly used region-based segmentation method. This paper describesthe origin of the CV model and its theory. However, CV model use the pixel value alonewithout the spatial location of the pixel. We propose a weighted CV model. The modelwhich adds the spatial relationships between pixels in the energy function can effectivelyimprove the accuracy of object segmentation and increase the curve smoothness.Canny operator is a common edge detection algorithms processing weak edge. But thecanny operator would detect many false boundaries which was not complete discontinuouswhen dealing with infrared images. In this paper, this problem has been solved by ratio cutedge extraction algorithm, which could decrease the false borders and detect a completetarget edge.Because of small target and complex background, we could detect target in the wholeimage. This article proposes a pre-processing method. The method uses the center-surrounddifference detecting the regions of interest (ROI).Then detect the Laws texture in the ROI,decreasing the affect of the complex background.At last we make a summary for the whole paper, then point out the aspects that need tobe considered and further research directions.