摘要
医学领域中,准确的检测癌细胞是非常重要的。目前识别细胞是否癌变,大都采用形态学、色度学等方法,使用阈值、小波变换等工具,对细胞边缘、细胞总面积、胞核面积、胞体形状、核内纹理等特征进行采集、分析、处理,最后对识别细胞是否病变。尽管在这些方面我们已经有了很多研究成果,由于细胞图像的复杂性与细胞组织的多样性,无法找到一种通用的算法来对各种细胞进行特征提取。基于此,本文采用圆谐傅里叶矩作为图像的特征,然后利用支持向量机将提取的图像矩进行优化,便可以得到一套准确、快捷的细胞识别算法。
The current systems of automatic image recognition are mainly based on morphology, colorimetry and other theories. In these systems, threshold value, wavelet transform are used to gather, analyze and conduct information on the edge, gross area of cells and nucleus, the shape of cyton and the texture inside nucleus, thus judging whether the cells are pathological or not. Though much achievement has been made in these aspects, due to the complexity of cell images and cell tissues, it is unable to recognize cell images in a unified way. We have to change methods to extract features of image according to the types of ceils, and the accuracy rate is not high. In this paper, Radial Har- monic Fourier moments （RHFMs） are chosen as the feature of cell images. SVM can be used to optimize the extracted image features and an accurate and rapid way for recognizing cells can be created.
出处
《太原科技大学学报》
2013年第5期372-377,共6页
Journal of Taiyuan University of Science and Technology
基金
太原科技大学软件工程专业培养模式的研究与实践(2011)
