基于Zernike矩特征提取的改进FCM手写体数字识别

Improved FCM Handwritten Digit Recognition Based on Zer-nike Moments Feature Extraction

摘要: 文中提出了采用Zernike矩进行特征提取，并通过改进FCM算法对手写体数字进行模糊聚类的一种方法。图像的Zernike矩具有旋转不变性，因此用它构造的特征空间能很好的反映图像的特性。模糊C均值聚类是用隶属度确定每个数据点属于某个聚类的程度的一种聚类算法，本文采取加权的模糊C均值聚类算法来进行分类。聚类的关键在于特征提取，在图像预处理的前提下，本文将Zernike矩进行降维处理，在不降低识别率的基础上，有效提高了识别速度。

Abstract: This paper proposes a method of Zernike moments feature extraction, which is through the improved FCM algorithm for fuzzy clustering of handwritten numerals. Zernike moments of the image have rotational invariance, so the feature space can well reflect the characteristics of the image. Fuzzy C-Means Clustering uses a degree of membership of each data point to determine the degree of belonging to a cluster, and this paper takes weighted fuzzy C-means clustering algorithm to classify the membership. Clustering lies in feature extraction. In the image pre-processing of the premise, this paper reduced the dimension of Zernike moments, and effectively improved the recognition speed without decreasing the recognition rate.