Up

A Fuzzy Neural Network Based Approach for CBIR

File Size:
1.76 MB
Volume:
Volume 3, Issue 8 (August, 2017)
Publication No:
IJTC201708007
Author:
Amanpreet Brar, Sukhpreet Kaur
Downloads:
5 x

Abstract:
Content Based Image Retrieval system based on medical images used to retrieves the similar type of medical image for a given input query image from large database. In this paper, we proposed a hybrid content based image retrieval system for medical images using fuzzy-feed forward back-propagation neural network technique. In this proposed work, Texture features such as contrast are extracted using GLCM (gray level co-occurrence matrix) Mean square energy, amplitude, standard deviation using Gabor filter respectively. Shape features are extracted using fuzzy edge detection .Combination of feature vectors is given as input to the neural network. Feed Forward back propagation neural network algorithm is used in training the neural network. Feed Forward algorithm propagates in forward direction to give output and back propagation algorithm calculates error in neural network. The fuzzy-(FFBP) neural network retrieves the similar images from database corresponding to query medical image .MATLAB 2010 b has been used to implement the proposed system. The results show that proposed method retrieves 100% precision and better recall values than other existing techniques.

Keywords:
Content- based image retrieval (CBIR) , Low-level descriptors , Fuzzy logic , Neural network , Feed –Forward , Back-propagation.