Comparative analysis of distance metrics for designing an effective content-based image retrieval system using colour and texture features

An enormous amount of information in the form of image and video are dispersed all over the world like any other data therefore, retrieval of a query image from a large database of images is an important undertaking in the area of computer vision and image processing. The traditional text-based approaches for searching images are slow and inefficient. Content-based image retrieval (CBIR) provides the solution for efficient retrieval of the image from these image databases. In this paper, an efficient CBIR system is proposed using various colour and texture features. Colour features such as Colour Moments and HSV Histogram and Texture Features like Local Binary Patterns (LBP) are used. Various distance metrics are analysed for retrieval and their performance is compared to get the best distance metric for better retrieval performance. From the experimental analyses on benchmark (WANG) database, it is observed that the City block distance performs consistently encouraging from other measures. Also this paper has introduced the combination of HSV and LBP histogram and evaluated the retrieval performance. The obtained results are very promising than other variants of colour and texture features.