Multiple Transform Domain Approaches for Content Based Image Retrieval System using Distance Measure Techniques
Content-based image retrieval (CBIR) has been implemented in real-time applications in the past few decades. In the retrieval system, the feature extraction method is a very important technique for a huge dataset. The dataset images are trained using various transforms, such as DWT, SWT, GWT, CT, and NSCT. The images are decomposed into various coefficients in terms of low- and high-frequency responses by using discrete wavelet and stationary wavelet transforms. These methods are not cover edge contours of the image. These are avoided in the Gabor, contourlet, and nonsubsampled contourlet transforms with various scaling and orientation properties. The proposed method was implemented with a nonsubsampled contourlet transform on the COREL dataset. The proposed method provided improved results for the performance measures such as recognition efficiency over the existing methods.