基于特征点的人体红外图像自动分割技术

目的医学红外人体图像区域分割是大规模医学红外图像处理的关键步骤。为快速有效地获取医学红外图像中的人体信息,本文提出一种在医学红外图像中自动提取并划分人体区域的方法。方法由红外热像仪在静室中采集人的裸体红外图像,然后通过对红外人体图像灰度分布特征分析而取得的阈值来获取人体区域,以人体横向距离(宽度)函数结合人体红外图像中的特殊方向亮带的识别,提取人体的特征点,并通过特征点对人体区域进行分割。结果对来自8人的72幅图像进行验证,其中64幅可以正确分割,证明该方法可以对直立姿势的红外人体图像进行自动区域分割与提取。结论该红外人体图像区域自动分割算法可为基于红外图像的疾病筛查及计算机辅助诊断提供技术基础。

Automatic segmentation on infrared images of human body based on feature points

Objective The segmentation of medical infrared image plays an important role in large?scale infrared image processing. In order to obtain the human body information quickly and effectively, this paper presents a method of segmenting and extracting human body regions in medical infrared images. Methods Infrared thermal imager collects infrared images of a naked person in a quiet room. The whole region of the human body is obtained by the threshold method. The threshold is obtained by analysis of the gray level distribution of medical infrared image,and the human body regions are segmented by the feature points of the human body which are extracted by the human body horizontal distance ( width ) function combined with the special direction bright band in the infrared image. Results There are 64 of 72 images from 8 persons divided by the method without bias. It proves that the method can segment and extract the human body regions from infrared images with human body in upright posture. Conclusions This segmentation algorithm for infrared image of human body can provide a technical basis for disease screening and computer?aided diagnosis based on infrared images.