Recovery of color loss due to specular illumination in Acne patient images

• Main Authors: Khan, J., Malik, A.S., Kamel, N., Das, S.C., Affandi, A.M.
• Format: Conference or Workshop Item
• Published: Institute of Electrical and Electronics Engineers Inc. 2014
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-84925645526&doi=10.1109%2fIECBES.2014.7047606&partnerID=40&md5=82f2d35b70695f296a4f3e5bceb8fac6; http://eprints.utp.edu.my/31780/

The occurrence of specular illumination phenomenon in digital photography depends on the nature of the surfaces, imaging geometry and lighting conditions. In the images of Acne patients, besides relative orientation of light source and objects, and lighting conditions; oily skin, pus (white oily liquid) inside Acne Vulgaris lesions and medication applied to face, also cause specular illumination. Such specular illumination washes away the color information of acne lesions and skin regions and makes the segmentation and analysis of digital color images very challenging. In this paper, a new image processing based technique is proposed for the detection of region affected with the specular illumination and recovery of color information of such regions. Energy of each pixel in the digital color is calculated and checked against a threshold to decide whether it is specularly illuminated or not. So as to recover the color information, a window of suitable size w is drawn around the specular pixel. Within the window, there can be three possible types of pixels; pixels belonging to skin, pixels belonging to Acne lesion and pixels which are affected with the specular illumination. Ignoring the pixels affected with specular illumination, the most probable color in the window is found by calculating histogram of the remaining pixels. This most probable color is assigned to the specular pixel around which the window was drawn. This technique was tested on the images of Acne patients taken at the Hospital Kuala Lumpur, Malaysia. The performance of the proposed technique is evaluated both subjectively and quantitatively. © 2014 IEEE.