An Experiment with OneCut and Active Contour Segmentation Techniques

Anuja Deshpande · International Journal for Research in Applied Science and Engineering Technology · 2018

This paper puts forth our observations from the experiments conducted on relatively newer interactive segmentation technique -One Cut based on Graph Cut and Active Contour segmentation technique based on Level Sets, using Matlab software on selected natural images.The images were selected such that those have complexities and pose segmentation challenge.The objective of this experiment is to understand and assess the effectiveness of these techniques on select natural images, which have complex image composition in terms of intensity, colour mix, indistinct object boundary, low contrast, etc.We have used Jaccard Index, Dice Coefficient and Hausdorrf Distance as measures to assess the accuracy, besides visual assessment.The segmented images were compared with ground truth and then subjected to these accuracy measures.We have found OneCut to be much more effective than Active Contour.

Read the paper · More papers on PaperTik