Robust Image Detection By Using Neural Network

Lingeswaran , Subramaniam (2005) Robust Image Detection By Using Neural Network. Project Report. UTeM, Melaka, Malaysia. (Submitted)

	PDF (24 Pages) Robust_Image_Detection_By_Using_Neural_Network_Lingeswaran_Subramaniam_QA76.87.L56_2005_2.pdf - Submitted Version Download (4MB)
	PDF (Full Text) Robust_Image_Detection_By_Using_Neural_Network_lingeswaran_Subramaniam_QA76.87.L56_2005.pdf - Submitted Version Restricted to Registered users only Download (27MB)

Abstract

This project will present an approach to image detection based on fast neural network methodology combined with a new algorithm. Cross- co-relation of neural network weights and reformulating the neural activities in hidden layer to enable the use of Fourier Transformation to speed up the time detection and other related parameters. In order to satisfy the requirements, for fast and robust detection of direction of an object on controlling the system, the neural network was adopted to the imaging processing system. The neural network modules got robust performance in detecting edges and directions of objects. It was important for doubling resolution of object direction to adjust original performance of the neural network for direction detection. In order to improve performance of direction detection of an object piled on the other one, trimming with edge pattern's embedded in random background noise was found to be essential. The image processing system can be operating very fast and robustly if the proposed neural network modules are composed by software programmers.

Actions (login required)

View Item