Image Classification Techniques

Abstract

Image classiffication techniques are used to classify di erent features available in the image. Image classi cation is the process of grouping image pixels into categories or classes to produce a Thematic representation. The objective of image classi cation is to identify the features occurring in an image in terms of the object or type of land cover. Image classi cation methods divide an image into regions which have same properties. These methods are applied in many areas such as medical imaging, object identi cation in satellite images, tra c control systems, brake light detection, machine vision, face recognition, ngerprint recognition, etc. Image classi cation techniques mainly divided in to two categories: Supervised im- age classi cation techniques and Unsupervised image classi cation techniques. Di er- ent supervised image classi cation techniques such as maximum likelihood classi er, arti cial neural network is implemented and analyzed. Di erent unsupervised image classi cation techniques Such as k-means algorithm , EM algorithm is implemented and analyzed. Accuracy of Supervised image classi cation techniques is depend upon selecting training pixels. Unsupervised image classi cation is used in real world when image does not have much information about data. Accuracy of Unsupervised im- age classi cation depend upon data itself for de nition of classes. To overcome this, wavelet based k-means algorithm is proposed, which is supervised image classi ca- tion technique. It has been observed form the simulation results that wavelet based k-means algorithm by supervised image classi cation gives better results.