基于多聚焦红外图像的温度测量及三维重构  

Temperature Measurement and 3D Reconstruction Based on Multifocus Thermal Images

作　　者：刘中意[1,2] 张春[2] 胡志宇[1,2,3] 

机构地区：[1]上海大学机电工程与自动化学院,上海200072 [2]上海大学纳微能源研究所,上海200444 [3]上海交通大学微米/纳米加工技术国家重点实验室,上海200240

出　　处：《红外技术》2015年第6期449-456,共8页Infrared Technology

基　　金：上海市科委科技攻关项目;编号:11DZ1111200;云南省科技厅省院省校科技合作专项;编号:2010AD003

摘　　要：由于红外镜头景深小,测量表面超过景深范围的样品不能得到所有位置均清晰聚焦的红外图像。提出了基于多聚焦图像融合的曲面温度测量及被测物体三维重构的方法。该方法采集同一视场不同聚焦深度的红外图像,利用多聚焦图像融合算法,获取所有位置均清晰聚焦的红外图像,即准确的红外信息。再查温度标定表,根据灰度值可得准确的温度值。最后,利用多聚焦图像系列附带的深度信息重构出表面三维高度图。实验表明,应用此方法可以准确测量物体表面温度信息,三维重构图与全聚焦图像形成信息互补。Because of the short depth of field of thermal camera, it is hard to get a thermal image of all-focus, especially the objects whose surface fluctuation is out of the depth of field. Points which are not focused in thermal image can＇t reflect the temperature information accurately. This paper proposes a method based on multifocus image fusion to resolve the problem. The method gets an all-focus image from multifocus images. The method improves the object temperature measurement error up to 0.68 degrees Celsius （average sharing to every pixel of thermal image） in working chip and 0.11 degrees Celsius in the skin of thumb. By the way, 3D surface reconstructed by depth form focus is achieved to give more information about test objects.

关 键 词：红外热像 多聚焦图像融合 温度测量 三维重构 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]