Performance Analysis of Fuzzy C-Means Clustering using Multichannel Decoded Local Binary Pattern

摘要

The construction of large database with thousands of data storage and image acquisitions have been facilitated with developments. Suitable information system requires proper handling of these datasets in efficient manner. ContentBased Image Retrieval (CBIR) is commonly used system to handle these datasets and on the basis of image substance the images that are related to the user given query for which the CBIR extracts the images from large image databases. The goal of feature extraction is to obtain the most relevant information from the original data and represent that information in a lower dimensionality space. Local Binary Pattern (LBP) based descriptors have been used for the purpose of image feature description. Local binary pattern has widely increased the popularity due to its simplicity and effectiveness in several applications. We used adder decoder based two schemas for the mixture of the LBPs from over one channel. Finally, Calculate feature vector to form a single feature vector. Clustering the image using Fuzzy Cmeans clustering under semisupervised framework.The experiments square measure executed over six benchmark color texture image databases. The performance of the proposed descriptors improved for three input channels and also in the RGB color space. The performance of mdLBP is also superior to nonLBP descriptors. It is pointed out that mdLBP outperforms the stateoftheart descriptors over large databases.