Pattern recognition using neural network : user defined imaged classifier based neural network using selected feature

Shirley, anak George. (2005) Pattern recognition using neural network : user defined imaged classifier based neural network using selected feature. [Final Year Project Report] (Unpublished)

Preview	PDF Pattern recognition using neural network... -(24 pages).pdf Download (8MB) | Preview
	PDF (Please get the password by email to repository@unimas.my , or call ext: 082-583914/3973/3933) Shirley Anak George.pdf Restricted to Registered users only Download (26MB)

Abstract

This study is about pattern recognition using Artificial Neural Networks (ANNs). The purpose of study is to train the Multi-layer perceptron (MLP) to decide whether an image contains desired object or not. The analysis for MLP is important to investigate its capability in capturing the images pattern as well as the goodness of selected feature used in this project. Preprocessing of images is done to find the edge of the object in an image. The edge features are processed before feeding to the MLP.The preprocessing is implemented in C, C++ language and OpenCV. Dataset that used in this project is collected from the researcher's collection. Experiment is carried out for the network using NevProp3 Neural Network simulation tool. The study shows that the MLP network is suitable to classify the images based on selected feature.The best classification rate is 72.7%

Actions (For repository members only: login required)

View Item