Image Segmentation and Extraction Regions Based on Single Seed Region Growing and Kmeans Algorithms

Abstract

In this article, region-growing image segmentation and K-means region extraction were the two techniques utilized to segment and extract regions from an image. As the region grows, determine the position of the chosen pixel, compare it to its neighbors, and label the selected region if the pixel falls inside the range of the threshold value that has been chosen. Continue this process for each subsequent pixel until there are no more pixels inside the threshold range. After extracting the regions and associated edge maps, which included the minimum and maximum intensity values, the number of pixels in the segmented regions was calculated. Convert the image from RGB color space to L*a*b* color space for K-Means clustering segmentation. Two color components (a*, which stands for red and green value) and lightness (L*) utilizing K-Means clustering, automatically divide the colors in the a*b* color space into three clusters by calculating the distance between them using the Euclidean Distance Metric. Then, use the cluster index results from K-Means to label each pixel in the image. then, using pixel labels and the cluster number that only includes the ROI among the three clusters, build an RGB label. Lastly, locate the divided image. Tests have been conducted on a variety of medical photos using both methods. The outcomes demonstrate the effectiveness and viability of the suggested ways by demonstrating how quickly and precisely the methods can extract regions of interest from both grayscale and color images.