|International Computer Science and Engineering Society (ICSES)|
ICSES Transactions on Image Processing and Pattern Recognition
Vol. 4, No. 4, Dec. 2018
Image Segmentation Based on Graph Theory and Threshold | Book Chapter of Image Segmentation: A Guide to Image Mining
Highlights and Novelties
This paper presents an image segmentation technique using discreet tools from graph theory. The image segmentation incorporating graph theoretic methods make the formulation of the problem suppler and the computation more ingenious. In our proposed method, the problem is modeled by partitioning a graph into several sub-graphs; in such a way that each of the subgraphs represents an eloquent region of the image. The segmentation is performed in a spatially discrete space by the efficient tools from graph theory. After the brief literature review, we have formulated the problem using graph representation of image and the threshold function. The borders between the different regions in an image are identified as per the segmentation criteria and, later, the partitioned regions are branded with random colors. In our approach, in order to make the segmentation fast, the image is preprocessed by DWT and coherence filter before performing the segmentation. We have carried out the experiments on numerous natural images available from Berkeley Image Database as well as synthetic images taken from online resources. The images are preprocessed using the wavelets of Haar, DB2, DB4, DB6 and DB8. In order to evaluate and compare the results, we have used the performance evaluation parameters like Performance Ratio, execution time, PSNR, Precision and Recall and found that the obtained results are promising.
© Copyright was transferred to International Computer Science and Engineering Society (ICSES) by all the Authors.
Cite this manuscript as
Vikramsingh Parihar, "Image Segmentation Based on Graph Theory and Threshold," in Image Segmentation: A Guide to Image Mining, 1st ed., ITIPPR: ICSES, 2018, pp. 61-82.
For External Scientific Databeses
Written by: Admin | Link ...