ADAPTIVE METHOD FOR AUTOMATIC IDENTIFICATION AND REMOVAL OF SPECULAR REFLECTIONS FROM AN IMAGE

摘要

Specular reflections (SR) are introduced in an image due to reflection from surface. It affects the original image in terms of its visualization quality. It€™s a noise and disturbs the identification of image contents and information. The SR in medical images is a real problem in computer aided diagnosis (CAD) systems. The intension of such CAD system is to automatically identify the medical image using various attributes and comment about the abnormality with respect to specific characteristics. Hence, detection and elimination of SR is a must before the actual image analysis. The un-treated SR pixels can affect the analysis in terms of feature extraction, training, classification etc.SR effect is common in all types of images on account of the reflection from surface. Motivated with this aspect, a completely adaptive method is proposed herewith, which will automatically enhance the quality of input image if required, and detect and remove SR pixels occurring in it.