【摘要】 目的手工选取关节区域的诊断方式增加了关节类疾病诊断的主观性和工作量。针对这一问题本文提出一种自动定位红外图像中人手关节的方法。方法根据人手的解剖学结构和手部特征点的位置,提出自动提取手关节区域的3种模型,即四边形模型、椭圆形模型和圆形模型,其中手部特征点包括目前手部图像研究中常用的11个特征点,以及本文新提出的2个辅助特征点。最后,通过一种基准图像比较方法验证本文方法的有效性。结果使用该方法能够准确找到关节中心的位置,在红外图像中自动定位出手部关节区域。结论该关节区域自动提取算法有助于关节定位及疾病诊断。更多还原

【Abstract】 Objective At present,joint areas still need to be located manually in the diagnosis of joint diseases,which is subjective and time consuming.To solve this problem,this paper presents a method for automated extraction of metacarpophalangeal(MCP) joints in infrared images of human hands.Methods Based on the averaged anatomical structure and the 13 landmarks of hand,three models:rectangle model,ellipse model and circle model are proposed.The 13 landmarks used in this paper include 11 common accepted landmarks in hand image research and two assistant landmarks proposed here.In addition,a new method of comparing with reference images is proposed to evaluate the accuracy and validity of the three models.Results The algorithm can help us to locate the centers of the joints and extract the joint areas accurately in infrared images.Conclusions The algorithm assists the localization of hand joints and is helpful to the diagnosis and therapy of joint diseases.更多还原