What is it about?

In computer vision, image segmentation is the process of partitioning a digital image into multiple segments (sets of pixels, also known as superpixels). The goal of segmentation is to simplify and/or change the representation of an image into something that is more meaningful and easier to analyze.

Featured Image

Why is it important?

Our method differs from the SSNCut with the following respects: (1) The method of generating superpixels is different, we use mean shift, and SSNCut uses SLIC. (2) The method of calculating the weights is different, we use Bhattacharyya distance to compute the similarity of the two regions, while SSNCut uses the globalized probability of boundary (gpb). Using the Bhattacharyya distance to define the edge weight of the graph is more efficient than gPb. (3)We use a new must-link and cannot-link constraint. (4) The segmentation method is different, we use the improved NCuts calculation. Compared to SSNcut, the calculation is more simple and effective.


The proposed method takes the advantages of the mean shift segmentation method and the NCuts grouping method.We have extend the NCuts algorithm for image segmentation and clustering to enable cannot-link and must-link constraints. Through various experiments on real images, the experimental results demonstrated that our approach generates better segmentation results and produce high quality segmentations, and the proposed method requires significantly lower computational complexity.

Shaojun Qu
Hunan Normal University

Read the Original

This page is a summary of: Supervised Image segmentation based on Superpixel and Improved Normalized Cuts, IET Image Processing, August 2019, the Institution of Engineering and Technology (the IET), DOI: 10.1049/iet-ipr.2018.6241.
You can read the full text:



The following have contributed to this page