Enhancing Texture Analysis by Color Shading Correction

The automated analysis and classification of textured regions in color images is an active field of research in image processing and pattern recognition. However, most of the published algorithms require ideal imaging conditions with respect to homogenous illumination conditions or calibrated recording devices. In this work we propose a YUVbased shading correction technique applicable to color images. Experiments conducted on images with synthesized shading show feasibility of the approach. Moreover the application of our algorithm to endoscopic images of the esophagus show significant gains in classification performance.