A Compact Auto Color Correlation using Binary Coding Stream for Image Retrieval

Auto color correlation (ACC) technique is proposed to enhance the availability of image content to capture local spatial correlation among different colors rather than using color correlogram technique. An ACC technique can reduce the size of color correlogram from O(md) to O(3md). However, it is still not applicable for query purposes in a large image database and especially in a real time image processing. This paper presents an application of a well studied image coding technique, namely block truncation coding (BTC), to reduce the storage space required and to increase the speed of retrieval for ACC algorithm. When an ACC is represented by using a binary matrix (BTC), it does not reduce the number of bins. Therefore, a decimal conversion of the binary stream in each row of the matrix is presented. The experimental results obtained from the two different techniques, Binary Coding Stream Conversion (BCSC) and ACC techniques, are investigated Key-Words: Retrieval, Auto Color Correlation, Block Truncation Coding, Image Processing, Color Descriptor