Study of segmentation method based on infrared images and visible-light images

Abstract

A technique based on infrared images acquisition from infrared thermal imaging technology combined visible-light images for image segmentation is introduced. Under the same hardware (infrared camera) acquisition devices captured the same point of the visible-light images and infrared images. The infrared image is used LOG operator to be objective edge. Then gets the original boundary to the larger lattice re-sampling, purpose to eliminate affected with infrared images by noise and ignores the breakpoints edge, the boundary chain code can be shorter. The boundary pixel record with 8 directions chain code. Use the chain code with the visible-light image from the same point of view for image segmentation. Thus, it can get the region of interest (ROI). Use to characteristics of infrared images low-resolution and less information, but sharp change on the edge. Ignore deal on the internal details of the object-region and focus to object-boundary. Use to characteristics of visible-light images color and texture information. Experimental results demonstrate that the Infrared images and visible-light images combined for image segmentation is effective and efficient.