clustering based algorithms on Images Of different resolutions for better segmentation and object detection

Abstract

In computer vision, image segmentation is the process of partitioning a digital image into multiple newlinesegments according to sets of pixels. Image segmentation is typically used to locate objects newlineand boundaries (lines, curves, etc.) in images. The result of image segmentation is a set of segments that newlinecollectively cover the entire image, or a set of contours extracted from the image. Many researchers have used newlinedifferent types of techniques for analyzing the image. For analyzing any image, clustering takes an important newlinerole. Clustering is the task of grouping a set of objects in such a way that objects in the same group (called newlinea cluster) are more similar (in some sense or another) to each other than to those in other groups (clusters). It newlineis a main task of exploratory data mining, and a common technique for statistical data analysis, used in many newlinefields, including machine learning, pattern recognition, image recognition, image analysis, information newlineretrieval, bioinformatics, data compression and computer graphics. newlineMany clustering algorithms are used for analyzing an image. In this thesis we have used two newlinealgorithms k-means clustering and Adaptive k-means clustering. k-means clustering algorithm belongs to newlinePartitional clustering. It is widely used in image processing specially in image segmentation. k-means newlineclustering segments the images but not gives satisfactory results of desire objects. newlineTo overcome the problems we used Adaptive k-means clustering algorithm. Both the algorithms newlineimplemented on images of different resolutions and compared the performance of the segmented images. The newlineperformances are compared based on some parameters. The parameters are PSNR, RMSE, SSIM, Elapsed newlinetime period. From each and every comparison we have concluded that the Adaptive k-means clustering gives newlinebetter result as compared to k-means clustering. By using these techniques we can isolate any target object newlinefrom medical images specially detection of cancerous data. Both the algorithms also implemented in other newlinefields like telecommunicat