早期癌变组织的图像识别方法

摘要

研究早期癌变组织的准确识别问题,由于人体癌变组织识别是对癌变部位与正常组织部位的像素灰度、梯度等特征差异进行识别,而在组织发生早期癌症病变时,癌变部位的图像灰度像素特征与普通组织部位的像素特征差异很小,传统的方法不能准确地识别人体早期癌变组织.为了解决上述问题,提出了根据病变部位像素密度的早期癌变组织识别方法.通过运用WORD像素理论,计算病变部位的像素数目,采用相关算法求得病变像素的像素密度,克服单一依靠像素灰度、梯度等直观特征进行识别的缺陷.仿真结果表明,改进方法能够有效的识别早期癌变组织,取得了比较好的效果.%Traditional human cancerous tissue recognition is based on the differences of image pixel gradient of cancerous tissue, and normal tissue, and cannot exactly recognise the early cancerous tissue in human body. In order to solve this problem, this paper proposed the early identification method of cancerous tissue based on the pixel densi-ty at lesion area. By using the Word pixel theory, the pixels number of pathological changes was calculated through die related algorithm for the pixel density of the lesions area. The experiment results show that the method can effec-tively identify early cancerous tissue, and achieve good results.