Image Segmentation using Histogram Clustering and Nonuniform Quantization technique
Seong Jin Gang, Seong Rak Gwon, Yeong Jo Lee, Chang Eon Gang · 1999
In this paper, we propose an image segmentation algorithm based on histogram clustering and nonuniform quantization, which is applicable to images with very complicated histograms. Conventional histogram-based image segmentation techniques are commonly used for segmenting simple images with the bi-modal histogram into segmentation techniquies are commonly used for segmenting simple images with the bi-modal histogram into distinguishable objects and background. However, it is difficult to apply those techniques to images with complicated multi-modal histogram. Accordingly, in this paper, we propose an image segmentation algorithm based on histogram clustering and nonuniform quantization. In the proposed algorithm, an input image is simplified using morphological filters, and the histogram is generated form flat regions in the simplified image.The segmented image is obtained by quantizing the simplified image with nonuniform quantization step which are determined by clustering the multi-modal histograms. Finally, region merging is carried out to eliminate tiny segmented regions from segmented image. Experimental results show that the proposed histogram based segmentation algorithm can be an alternative to the conventional segmentation algorithm.