基于改进FCM聚类算法的维吾尔文字图像分割

Improved FCM Clustering Algorithm Uighur Text Image Segmentation

研究维吾尔文字图像分割问题,针对传统的FCM聚类算法对维吾尔文字符图像分割时相邻域的信息未能考虑,故容易造成其对维吾尔文字符图像分割时的缺陷和干扰问题,为解决上述问题,提出了一种改进型FCM聚类算法的维吾尔文字符图像分割的方法。首先通过多尺度图像锥建立了其聚类算法,减少维吾尔文字符图像分割时的数据大小,进而一步降低了维吾尔文字符图像分割本身的计算量,然后通过空间信息的引入,使干扰噪声信息被屏蔽,从而提高维吾尔文字图像聚类分割的抗干扰能力。仿真结果表明,算法更容易提高维吾尔文字图像分割效果,准确分割出维吾尔文字区域,提高了识别精度,优于一般的FCM聚类算法。

Study the Uygur text image segmentation. When the information can not be considered adjacent do- main, the traditional FCM clustering algorithm for image segmentation Uygur character likely causes Uygur character image segmentation defects and interference. In view of this, the paper proposed a modified FCM clustering algorithm for Uygur character image segmentation. This method first established the multi-scale image of its cone clustering al- gorithm to reduce the Uygur characteristics of image segmentation and the data size, and thus further to reduce the calculation of Uygur character image segmentation own. By introduction of its spatial information, the interference noise information was shielded, thereby improving the anti-jamming capability of Uygur text image clustering segmen- tation. Simulation results show that the algorithm is easier to improve the Uygur character image segmentation and its recognition accuracy.