Paper: A Novel Color-texture Descriptor Based on Local Histograms for Image Segmentation
In this paper, we propose a novel color-texture image segmentation method based on local histograms.
Starting with clustering-based color quantization, we extract a sufficient number of representative colors. For each pixel, through counting the number of pixels with each representative color within a circular neighborhood, a local histogram is obtained.
After the circular neighborhood is extended to several scales, a local histogram with an appropriate scale is adopted as a color-texture descriptor at the corresponding pixel for image segmentation.
Further, we correct the color-texture features near boundaries and obtain a initial segmentation by a clustering method with the color-texture descriptors.
Finally, in order to obtain a better segmentation result, we merge the over segmented regions guided by the obtained boundaries.
Experiments are performed on both synthetic and natural color-texture images, and the results show that our proposed method performs much better compared with state-of-the-art methods on image segmentation, particularly in textured areas.
A novel color texture descriptor is designed to obtain per-pixel texture feature within a circular neighborhood with an adaptive range. Simultaneously, the texture boundaries are extracted by our proposed method. Finally, the image segmentation is performed by clustering the obtained texture features and merging small regions under the guidance of the texture boundaries.
Liu, Yang; Liu, Guangda; Liu, Changying; Sun, Changming. A Novel Color-texture Descriptor Based on Local Histograms for Image Segmentation. IEEE Access. 2019; 7:160683-160695. doi.org/10.1109/ACCESS.2019.2951228
Download the full paper here.
For more information, contact us.