| 摘 要: 针对简单线性迭代聚类(Simple Linear Iterative Clustering,SLIC)算法对不同图像自适应性差的问题,提出了一种基于皮尔森相关系数的自适应SLIC超像素图像分割算法。首先,通过量化非间隔进行图像预处理,并计算颜色熵作为图像复杂度,从而确定所需分割的超像素个数。其次,利用皮尔森相关系数作为相似性度量函数。最后,通过纹理特征对类内异常点进行滤除,确保种子点更新的准确性。实验结果表明,在超像素个数相同的情况下,基于皮尔森相关系数的自适应SLIC超像素图像分割算法相比主流超像素分割算法,可以获得更高的边缘命中率以及更低的欠分割率,性能优于LSC(Linear Spectral Clustering)、SLIC和SLIC0(Simple Linear Iterative Clustering Zero)算法。 |
| 关键词: SLIC算法;自适应性;皮尔森相关系数;超像素图像分割;图像复杂度 |
|
中图分类号: TP391
文献标识码: A
|
| 基金项目: 湖南省自然科学基金项目(2021JJ50049)；湖南省教育厅重点项目(21A0607)；湖南省自然科学基金项目(2022JJ50077). |
|
| Adaptive SLIC Superpixel Segmentation Algorithm Based on Pearson Correlation Coefficient |
|
LIANG Haohan, WANG Zhiqiang, CUI Peng
|
(Department of Computer Science and Technology, Harbin University of Science and Technology, Harbin 150000, China)
1448200309@qq.com; 15216806589@163.com; 641397290@qq.com
|
| Abstract: This paper proposes an adaptive Simple Linear Iterative Clustering ( SLIC) superpixel image segmentation algorithm based on Pearson correlation coefficient to address the issue of poor adaptability of SLIC algorithm to different images. Firstly, image preprocessing is performed by quantifying non-intervals, and color entropy is calculated as the image complexity to determine the number of superpixels required for segmentation. Secondly, Pearson correlation coefficient is utilized as the similarity measurement function. Finally, texture features are used to filter out intra-class outliers, ensuring the accuracy of seed point updates. Experimental results demonstrate that, under the same number of superpixels, the proposed algorithm can achieve higher edge hit rate and lower under-segmentation rate, compared to the mainstream superpixel segmentation algorithms on the BSD500 dataset. The performance of the proposed algorithm surpasses that of LSC (Linear Spectral Clustering), SLIC, and SLIC0 (Simple Linear Iterative Clustering Zero) algorithms. |
| Keywords: SLIC algorithm; adaptability; Pearson correlation coefficient; superpixel image segmentation;image complexity |
