Description

Content Based Image Retrieval is the process of identification of similar images present in the database. A Feature based approach is employed for the identification of the similar images from the dataset. The features were extracted from the images based on Dot Diffused Block truncation coding (DDBTC) compressed image. Color Histogram features and Bit Pattern features were extracted from the images. The similarity between the features were measured in-terms of distance measurements. Five similarity measures were used for the identification of similarities in the images. The performance of the process is measured in-terms of Precision, Recall, Average Precision Rate (APR) and Average Recall Rate (ARR). The input color images were collected from the image retrieval database. The Red, Green and Blue color channels were separated from the input image. The color quantizers were extracted from the image minimum and maximum color space of the images. The extracted quantizers were used for the construction of the minimum and the maximum quantizers. Then Dot Diffused Block truncation coding (DDBTC) is performed for the encoding of the images. The Bit Pattern Image is generated with the help of the encoding process. The Color Histogram features were extracted from the minimum and maximum quantizers. Decoding process can be done based on the reversal process of the Bit Pattern image. The quantized images were divided into blocks. The diffusion pattern is identified in the particular block. The diffusion pattern is then quantized using LBGVQ quantization and the histogram values were calculated which acts as the color histogram features of the input image is. In the Bit Pattern Images the same process applied for the quantized image is applied and the Bit Pattern Features were extracted from the images. The similarity between the features were calculated based on the L1 distance, L2 distance, x2 distance, Fu distance and Modified Canberra distance.