A Novel Method for Eye Contour Extraction from Blurred Infrared Images

摘要

Thermal image taken by infrared thermography can reflect distribution of skin temperature and indirectly reflect the physiological status. However, the acquisition of eye temperature with manual or semi-automated method will make computer-aided diagnosis system less automated. Moreover, infrared image is characterized by low resolution and blur, which makes it more difficult to extract eye contour with traditional active contour models. This paper proposes a novel active contour model, which is able to deal with blurred infrared image of eyes. And it optimizes iterated procedure by calculating energy changes directly. By this way, the proposed algorithm allows fast convergence to solutions. Experimental results show that compared with other algorithms, the algorithm proposed in this paper performs better.