The occurrence of highlights in digital photography is an unwanted phenomenon in computer vision and image analysis. It overshadows the intrinsic color information of objects in a scene and makes computer vision algorithms to produce erroneous results. In this paper, two techniques are proposed: one for removing highlights consisting of both diffuse and specular components and the other one for removing highlights due to specular components only. The former is based on the modified Torrance-Sparrow model of specular reflection while the later technique combines Torrance-Sparrow model with local statistical analysis and recovers the color of highlights by interpolation. The results are validated on both synthesized and real images and compared with several contemporary highlights removal techniques. The comparison shows that the performance of the proposed techniques is comparable or slightly better than the previously proposed techniques. © 2019 University of Kuwait. All rights reserved.