Research On 3D Model Of Temperature Distribution Based On Thermal Infrared Image

Infrared images are widely used in medical,agriculture,military and other domains.Comparing to color images,thermal images eliminate the illumination problems and apply to night vision.Due to these advantages,if the spatial information of the temperature distribution is provided to form three-dimensional(3-D)thermography,more accurate temperature information could be acquired and infrared thermography can be extended to more applications.Based on binocular stereo technology,a reconstruction system is built to acquire three dimension temperature distribution.Key techniques,including system calibration,stereo correspondence and thermal images matching to 3D model are explored and researched in depth.The prime research task content is as follows:1.In stereo system calibration,Zhang plane calibration method is adopted to accomplish color camera calibration.Conventional calibration techniques cannot be directly applied to IR camera because it is sensitive only to thermal information.Hence,a calibration patterns is designed,and corresponding calibration method is proposed to calibrate the infrared camera accurately.2.Stereo matching algorithm based on the traditional belief propagation has been discussed in depth.The global energy function of the traditional belief propagation algorithm cannot effectively balance data items and energy items.In order to effect inhibit the phenomenon of parallax empty and reduce the false matching rate,two adjustment coefficients were presented to adjust the energy function.3.To obtain the temperature distribution of the target surface,a mapping method is proposed based on spatial relationship between the color camera and the infrared camera.Points in 3D model are projected to the thermal image and return the temperature value to the 3D model.Then,a surface interpolation algorithm is adopted to acquire complete 3D temperature model.