International Conference on Advances in Computer and Information Technology - ACIT 2013
Author(s) : H. RANGANATHAN, P.S.RAMAPRABA
In this paper colposcopy cervical image classification based on colour histogram and K Nearest Neighbor (KNN) is presented. The classification is achieved by extracting colour histogram features from the cervix. To extract the colour histogram features, the colour space of the given image is converted from RGB to CIE colour space because of its perceptual uniformity. KNN classifier is used to classify the cervical images into normal and abnormal images. The performance with overall sensitivity of 94.71% and accuracy of 93.75 % is achieved using k-NN classifier. The performance is evaluated using 240 images collected from the hospital.