Digital image clustering and colour model selection in content-based image retrieval (CBIR) approach for biometric security image

Abstract

In this paper, we mainly discuss about the Digital Image Clustering in Content-Based Image Retrieval (CBIR) approach for biometric security. The main focus in this paper is to collect and gather as much as information about the technique for feature extraction in order to reduce the semantic gap problem in Content-Based Image Retrieval (CBIR) to be implemented in biometric (face recognition) process. Several general features of Content-Based Image Retrieval (CBIR) have been explained and discussed in this paper such as colour features, texture features, and shape features. From the preliminary study conducted earlier, colour features has been selected for the implementation of features extraction. In terms of biometric image features extraction implementation, image features of facial image will be extracted and stored in image feature database. Next, the query image will be processed in the same way as images in the database. Then, the matching is carried out on the feature database. On top of that, the common algorithm of colour features has been discussed as well to select the suitable colour model to be used. Based on the finding, the Red Green Blue (RGB) has been chosen as the selected colour model. The summarization regarding performance study of Content-Based Image Retrieval (CBIR) algorithm based on RGB colour features selection has been included in this paper. The purpose of that performance study is to analyze the advantages and limitations of implementing the RGB colour model in the Content-Based Image Retrieval (CBIR) technique. Other than selection of the general features of Content-Based Image Retrieval (CBIR), the selection and nomination of the clustering algorithm is very important in order to provide a greater Content-Based Image Retrieval (CBIR) system. Several common algorithms such as K-means, Isodata, and KHM from the Partitional Clustering Algorithm has been discussed in detail in this paper.