Image understanding using decision tree based machine learning

Abstract

Image Understanding, a discipline that concerns the interpretation of an image and analysis of the image to give a decision about the image and the actions represented in it. Decision tree is a tree based classification, widely used in data mining, which classifies the input data set into predefined classes. Decision tree approach is used here to train the image understanding system to perform supervised machine learning. The various low level characteristic features (color, shape, texture) of the image form the various attributes of the decision tree among others. This paper presents the application of the decision tree approach for image understanding. It also discusses an algorithm to calculate the relative distance between the retrieved results, as a sub process required in the proposed approach. The paper describes the production rules required to generate the decision tree. An example study is used to describe the image understanding process in a descriptive manner.