Histogram-based image segmentation using variational mode decomposition and correlation coefficients

作者：Hao, Duo; Li, Qiuming; Li, Chengwei^*

来源：Signal, Image and Video Processing, 2017, 11(8): 1411-1418.

DOI：10.1007/s11760-017-1101-z

• Image segmentation
• Variational mode decomposition
• k-means
• histogram
• Correlation coefficient

Image segmentation is defined by separating scenes into different parts. The grayscale histogram-based method is a widely used method in image segmentation because of its low computational complexity. However, traditional grayscale histogram-based methods usually suffer from wrong segmentation or low structure because of the irregularities and sharp details on the grayscale histogram and the absence of spatial information. This study presents an accurate image segmentation method by using the variational mode decomposition (VMD) and correlation coefficient (CC). First, grayscale histogram is decomposed into certain band-limit modes by VMD to remove the adverse effects. Then, the candidate centroids of the clusters are obtained by searching the refined histogram both vertically and horizontally. The K-means method is used to generate the segmentation results. Finally, some over-segmented regions are merged following the CCs of spatial histogram, which is also refined by the VMD method. The proposed method was applied to four images. The obtained experimental results show that the proposed method can generate reasonable segmentation results than those of the existing image segmentation algorithms.

• 出版日期2017-11
• 单位哈尔滨工业大学