Eye Configuration as Identifier of Person Using Fractal Dimensions

Abstract

The need for fast image retrieval of persons in the ever growing amount of digital newlineimages is steadily growing. In this work we are primarily estimating the gender and newlineage of a person. We have collected data of persons of different age group and gender newlinein the form of images. newlineEyes expression communicates emotions, pain etc. and shows various human newlinebehavior. Each person was asked to give different facial expressions. We have newlineconsidered basic emotions in the form of rasas for gender and age identification of a newlineperson. Nine rasas such as adbhutam, bhayankaram, bibhatsam, hasyam, karunnam, newlineraudram, sringaram, viram and shantam are universally recognized expressions in newlineIndian classical dance. newlinePersons were divided in two genders male and female. Each gender was divided in newlinefour age groups 20-30, 31-40, 41-50, 51-60. Each person of different gender and age newlinegroup was asked to perform different rasas. newlineEye and eyebrow regions are cropped from the image. For cropping the interested newlineregion we have modified the existing method of valley field algorithm. In the existing newlinealgorithm two rectangles was taken for two eye. In the modified valley field method, newlinewe have joined the two rectangles and extended it up to eyebrows. By setting the newlinethreshold values, we identify the eye and eyebrow region that means interested area in newlinethe form of rectangle consist of eyes and eyebrows and not any other part of face. newlineThis is called cropped image. This cropped image is inputed in m file. File m is a newlineMatlab file in which source code is developed. newlineCorresponding to each person ten statistical parameter like Entropy, Mean, Area, newlinePerimeter, Histogram area, Variance, Kurtosis, Skewness Moment, Standard newlineDeviation are taken for each of nine rasas like Sringaram, Hasyam, Raudram, newlineKarunyam, Bibhatsam, Bhayankam, Viram, Adbhutam, Shantam. Values of each newlinestatistical parameter are calculated through program developed in MatLab11. This newlinewas the part of standardization. newlineMinimum and maximum values of each age group are calculated and graphs between newlinehistogram area,