Superpixel segmentation algorithm based on local network modularity increment

Abstract

Superpixel segmentation is a kind of image preprocessing technology and a popular research direction in image processing. The purpose of superpixel segmentation is to reduce the complexity of image processing. The most widely applied Simple Linear Iterative Clustering (SLIC) superpixel segmentation algorithm has high operating efficiency. However, under-segmentation is prone to occur when the number of given superpixel regions is too small. In order to improve the segmentation accuracy, the superpixel segmentation algorithm based on local network modularity increment (LocalNet) from the perspective of network community detection is proposed here. The adjacency network is constructed according to the colour similarity of image pixels, the local community centre is found by the degree of network nodes, and the local network structure of the community is constructed. The modularity increment is employed as the boundary constraint to improve the segmentation accuracy of superpixel segmentation. Through the experimental comparison with the SLIC algorithm, its improved algorithm, and the algorithm proposed in recent years, the results show that our LocalNet algorithm significantly improves in segmentation accuracy Furthermore, the segmentation effect has obvious advantages under the premise that the segmentation speed is not much different from that of the other five algorithms.

An Efficient Hyperspectral Image Classification Method: Inter-Class Difference Correction and Spatial Spectral Redundancy Removal

“…Superpixel segmentation involves placing similar pixels in the same superpixel block. The superpixel block has boundary information, and the adjacent superpixel blocks satisfy a seamless topological relationship [20]. In the HSIs, there is a high probability that adjacent objects in space are similar.…”