A simple and fast automatic face colour identification for reducing the search space

摘要

Accuracy is the most important objective of this missing child identification problem. If the database is huge and the age gap between the missing time and the time of identification is large then the accuracy is very low. The accuracy of detection increases with reduction of search space or the database size. Therefore there is a need for reducing the search space into an appropriate smaller search space where the possibility of getting information about the query image is high. In this paper we tried to reduce the search space of missing persons by classifying into three colour class. The search space is classified into three sub space according to the skin colour of the query image. We proposed two methods to divide the entire data space into Dark, Medium and Fair groups. The two proposed methods are based on histogram feature and statistical features. The proposed methods help to reduce the search space approximately upto two third of the actual search space.