Improving Performance of Image Retrieval Based on Fuzzy Colour Histograms by Using Hybrid Colour Model and Genetic Algorithm

Abstract

A hybrid colour model is a colour descriptor formed by combining channels from several different colour models. Although rarely used in computer graphics applications due to redundancy, hybrid colour models may be of interest for the Content‐Based Image Retrieval (CBIR). Best features of each colour model can be combined to obtain optimal retrieval performance. This paper evaluates several approaches to the construction of a hybrid colour model that is used to construct a fuzzy colour histogram of image as a compact feature for retrieval. By evaluating each channel separately, a colour model named HSY is proposed. Various parameters of fuzzy histogram are further improved using Genetic algorithm (GA). Using standard data sets and the Average Normalized Modified Retrieval Rank (ANMRR) as a metric for retrieval performance, it is shown that this novel approach can give an improved retrieval performance.A hybrid colour model is a colour descriptor formed by combining channels from several different colour models. Although rarely used in computer graphics applications due to redundancy, hybrid colour models may be of interest for the Content‐Based Image Retrieval (CBIR). Best features of each colour model can be combined to obtain optimal retrieval performance. This paper evaluates several approaches to the construction of a hybrid colour model that is used to construct a fuzzy colour histogram of image as a compact feature for retrieval. By evaluating each channel separately, a colour model named HSY is proposed. Various parameters of fuzzy histogram are further improved using Genetic algorithm (GA). Using standard data sets and the Average Normalized Modified Retrieval Rank (ANMRR) as a metric for retrieval performance, it is shown that this novel approach can give an improved retrieval performance.