Improved technique for object detection using Similarity Based Region Merging

I. Introduction Image Segmentation is a process which partitioned image into multiple unique regions, where region is set of similar pixels. If I is set of all image pixels, then by applying segmentation we get different-different unique regions like {R 1 ,R 2 ,R 3 ,..,R n } which when combined formed the image I. Many approaches have been proposed earlier which includes probabilistic graphical models[1], normalized cuts[2], graph-cut method [3] ,region growing[4] etc. .The automatic segmentation approaches may also fail even though they use prior information in segmentation process. The main reason behind this is the complexity of object segmentation of image in real applications. The complexity is due to the several reasons like shadow, low contrast areas, occlusion, cluttering and noise in the image. These reasons make the segmentation process quite difficult and challenging. The use of interactive image segmentation process is the solutions of such problems. In the interactive image segmentation the proper user's intervention is required for segmentation process. Because user gives the clue for segmenting the image in the process the results are more improved and satisfactory .Therefore Semi-automatic or interactive segmentation method is proposed, which use human expert knowledge as additional input and makes the segmentation problem more tractable. The goal of semi-interactive segmentation methods is to minimize the required user interaction time, while maintaining tight user control to guarantee the correctness of the results .In image segmentation image similarity measure play an important role. Region merging for object retrieval is an important task in many image processing applications. It has a wide application in area of crime prevention, intellectual properties, medical diagnosis, web searching etc. The object segmentation results are very much influence with how to groups sub regions. So, accurate object segmentation is possible if we combine both high level and low level priors effectively. To achieve this we introduce in our paper new approach object segmentation with region labeling and region growing.