Abstract: To create an efficient classification method using a multi structural analogy for images. Classification is a division of data into groups of similar objects. Representing the data by fewer clusters necessarily loses certain minor details, to minimize the minute error, come up with a theme is to increase and create an efficient classification technique. When we have a large number of data to be processed and it needs to be clustered for simplification. We have to use efficient approach to consume less time should be taken into consideration. This paper represents creating such methodology in keeping mind all the pros and cons occurred in clustering .We have come up with an idea, using multi features like image feature extraction, parallel computing and distance finding algorithms which bind with sum product tree will results in achieving an efficient classification.

Keywords: Parallel computing, K nearest neighbor, canny edge detection, sobel edge detection, sum product tree.