Reflection Removal of Tongue Image via Total Variation-Based Image Inpainting

With the progress in image processing and pattern recognition, computerized tongue diagnosis has received considerable recent research interest. In computerized tongue diagnosis, reflection is unavoidable in the acquisition of tongue images. Correct detection of the highlight area and removal of reflection in tongue images is requisite and would greatly affect the performance of tongue texture analysis. In this paper, we first present a highlight detection method based on the color model, and then adopt a total variation (TV)-based image inpainting method for reflection removal. Fast gradient method is applied to enforce fast inpainting with guaranteed convergence performance. Experimental results on simulation and real data show that the proposed method is effective in highlight area detection and reflection removal.