Improvisation of Content Based Image Retrieval Using Color Edge Detection with Various Gradient Filters and Slope Magnitude Method

Abstract

Content based image retrieval (CBIR) is the better approach for image retrieval than traditional methods. Images can be retrieved using color edge detection and discrete wavelet transforms [2]. The paper discusses extended version of the image retrieval methods based on canny edge detector and Haar wavelet transform with the shape features extracted using gradient operators (Sobel, Prewitt, Laplace, Frei-Chen) and slope magnitude technique. The proposed image retrieval technique is tested on image database with 350 images. The database consists of 7 categories of images. Ten images from each category are taken as a query to find precision values for proposed method. The similarity criteria used is Manhattan Distance for calculating similarity between stored image and query image. The average precision of all queries are computed and considered for performance analysis. In proposed algorithm, Frei-Chen and Sobel gives better performance for almost all image categories than existing image retrieval method which uses canny operator for edge detection.