Simple and efficient method for specularity removal in an image

Abstract

For dielectric inhomogeneous objects, the perceived reflections are the linear combinations of diffuse and specular reflection components. Specular reflection plays an important role in the fields of image analysis, pattern recognition, and scene synthesis. Several methods for the separation of the diffuse and the specular reflection components have been presented based on image segmentation or local interaction of neighboring pixels. We propose a simple and effective method for specularity removal in a single image on the level of each individual pixel. The chromaticity of diffuse reflection is approximately estimated by employing the concept of modified specular-free image, and the specular component is adjusted according to the criterion of smooth color transition along the boundary of diffuse and specular regions. Experimental results indicate that the proposed method is promising when compared with other state-of-the-art techniques, in both separation accuracy and running speed.

Removal of parasitic image due to metal specularity based on digital micromirror device camera

“…[1][2][3][4][5][6][7][8] However, when scattering glare is in the neighboring region, these techniques are no longer valid. There are many methods to remove various parasitic images.…”

Separation of reflection components by sparse non-negative matrix factorization

“…To cope with this difficulty, more recent methods attempt to utilize spatial information in the image [9,15,[10][11][12][13]. Instead of determining body colors first, most of these methods search for all the parameter values simultaneously through optimization.…”

“…Shen et al [11] proposed another specular-free image that is obtained by subtracting the minimum of the RGB values from them for each pixel. They proposed a simple separation method based on it and also on an incorporated prior that the body color changes smoothly around highlights.…”

“…A large number of studies have been conducted to develop a method for accurately and robustly separating reflection components in a single image [4][5][6][7][8][9][10][11][12][13]. Most of them assume the dichromatic reflection model, which states that the light reflected on an object surface is given by a linear sum of a specular component and a diffuse component [4].…”

“…More recent studies [9,15,[10][11][12][13] attempt to utilize spatial information to improve separation accuracy. To do so, they incorporate spatial priors such as the smoothness of the body colors and/or the specular reflections on the object surface.…”