Publication

Abstract : Wavelets play an important role in image processing, as these are used to derive necessary information from an image. This paper presents a novel approach for designing a new wavelet for each ear image through a ringprojection technique. With this technique, 2-D ear image dimensionality reduces to 1-D information, which is useful in designing the adaptive wavelet for each input image. The 1-D information pattern obtained through ring-projection is invariant to rotation and less memory is needed to store the data. The principle for designing a new wavelet is to minimize the error signal of given information and express the resulting information as smoothening signal under constraints leading to an adaptive wavelet. The WPUT ear image database is used to examine the superiority of wavelet designing technique; the database contains 4 to 8 images per subject. Ear image of each subject contains normal image, light variation, accessories or hair occlusion, etc., adaptive wavelet is designed for the normal ear image. By the designed adaptive wavelet, all the ear images of the same subject are decomposed up to second level, to extract ECR value. When hindrance is present in ear image, then variations of ECR value indicate the amount of useful information contained by each image after decomposition using wavelet.