Optimization of image retrieval by using HSV color space, Zernike moment & DWT technique

Abstract

Content Based Image Retrieval (CBIR) is the task of retrieving the images from the huge set of database on the basis of their own visual content. Content based image recovery is utilized for the programmed indexing and recovery of images depending on the contents of images called as the elements. This paper gives indicated way to utilize these primitive elements to recover the desired image. The procedure by which we acquire the provides image is CBIR. In the CBIR first color space of HSV is quantified to obtain the color histogram. Apply Color Correlogram for color features which are utilized for calculating distance between two different colors. Apply DWT (Discrete Wavelet transform) for surface elements. It extracts features in four blocks-low-high filter and combine features with standard and mean deviation values. Apply Gabor Filter for measuring orientation of texture features. Using Zernike moment, identify shapes of an image. Utilizing these parts an element grid is shaped. At that point this lattice is mapped with the normal for global color histogram and local color histogram, which are analyzed and looked at. In light of this standard, CBIR system uses color, surface and shape, fused elements to recover the desired image from the huge amount of database and subsequently gives more effectiveness or improvement in image recovery than the single component recovery system which means better image recovery results. Classify the data using radial basis kernel support vector machine. For the experiment, we used Wang Database of 1000 images. It gives accuracy around 70–85%.