Research On Enhancement Of Thermal Image Of Hand Based On The Retinex Theory

As a functional imaging technique, medical infrared thermal imaging technology can reflect people's health status with the thermal images, which represents the temperature distribution of human tissues with different gray values. It can be used to diagnose and prevent the diseases in clinic, and has a broad application prospect. The details in the infrared thermal image can't be easily identified because of many factors. So it is necessary to enhance the infrared thermal images. The image enhancement methods on infrared images are studied in the paper, especially the algorithms based on Retinex theory.Firstly, the enhancement algorithms based on gray transform is studied, including linear transformation, nonlinear transformation, histogram equalization, and histogram normalization. Some thermal imaging experiments on hand have been done, and the result shows that the processing effect is not very satisfactory. Secondly, algorithms based on Retinex theory are studied. The qualities of the infrared thermal images using the global-Retinex theory have been improved, but it is time consuming, some influences of gray values non-correlative with the pixels are introduced, which lead to the loss of some details and the instability of the dynamic range of the images. Furthermore, the local Retinex algorithms are studied, including the single-scale Retinex algorithm, Frankle-McCann algorithm, McCann99 algorithm and multi-scale Retinex algorithm, and applied to the infrared thermal image of hand. The results showed that using the improved multi-scale Retinex algorithm on the basis of color restoration can get the best effect, which the color is natural and rich, and the fidelity is increased.