| 融入类贡献抑制因子的灰度级模糊C均值图像分割 |
| |
| 引用本文： | 朱占龙,,,马艳玲,,,董建彬,,,郑一博,.融入类贡献抑制因子的灰度级模糊C均值图像分割[J].智能系统学报,2021,16(4):641-648. |
| |
| 作者姓名： | 朱占龙 马艳玲 董建彬 郑一博 |
| |
| 作者单位： | 1. 河北地质大学 信息工程学院，河北 石家庄 050031;2. 河北省光电信息与地球探测技术重点实验室，河北 石家庄 050031;3. 河北省智能传感物联技术工程技术研究中心，河北 石家庄 050031 |
| |
| 摘 要： | 基于灰度级模糊C均值图像分割算法具有分割速度快的优势。由于无损检测图像中背景类和目标类差异较大，该算法不能有效地将目标分割出来，故提出改进的基于灰度级的模糊C均值算法。构建了一种与类大小反向相关的类贡献抑制因子表达式，将之融入目标函数后能够降低较大类对目标函数的贡献，这可避免较小类的聚类中心受较大类的影响而靠近较大类的聚类中心。最小化新的目标函数可得新形式的隶属度和聚类中心表征形式。采用类大小差异较大的无损检测图像进行试验，结果显示本文算法得到的分割图像视觉效果良好，而且指标G_mean也更高，进一步提升了基于灰度级模糊C均值算法适应能力。
|
| 关 键 词： | 模糊C均值算法 图像分割 灰度级 空间信息 无损检测图像 去噪 聚类中心 目标函数 |
Gray level-based fuzzy C-means algorithm for image segmentation with inhibitors of cluster contribution |
| |
| ZHU Zhanlong,,,MA Yanling,,,DONG Jianbin,,,ZHENG Yibo,.Gray level-based fuzzy C-means algorithm for image segmentation with inhibitors of cluster contribution[J].CAAL Transactions on Intelligent Systems,2021,16(4):641-648. |
| |
| Authors: | ZHU Zhanlong MA Yanling DONG Jianbin ZHENG Yibo |
| |
| Affiliation: | 1. School of Information Engineering, Heibei GEO University, Shijiazhuang 050031, China;2. Hebei Key Laboratory of Optoelectronic Information and Geo-Detection Technology, Shijiazhuang 050031, China;3. Intelligent Sensor Network Engineering Research Center of Hebei Province, Shijiazhuang 050031, China |
| |
| Abstract: | The fuzzy C-means algorithm for image segmentation based on the gray level has the advantage of fast segmentation speed. However, this algorithm cannot effectively segment object pixels from a nondestructive testing (NDT) image because of the great difference between the background and the object region. Therefore, an improved fuzzy C-means algorithm based on gray level is proposed. First, an expression called inhibitors of cluster contribution, which is inversely related to the class size, is constructed. Incorporating this expression into the objective function reduces the contribution of the larger cluster to the objective function, which then avoids the influence of the larger cluster on the cluster center of the smaller cluster. Second, a new form of membership degree and cluster center representation can be obtained by minimizing the new objective function. Lastly, NDT images with a large difference in cluster size are used for testing. Results show that the segmentation image obtained by this algorithm has a better visual effect and the index G_mean is higher, thereby further improving the adaptability of the fuzzy C-means algorithm based on gray level. |
| |
| Keywords: | fuzzy C-means algorithm image segmentation gray levels spatial information non-destructive testing image denoising cluster center objective function |
|
| 点击此处可从《智能系统学报》浏览原始摘要信息 |
|
点击此处可从《智能系统学报》下载全文 |
|
