Temperature Measurement And 3D Reconstruction Based On Multi-focus Thermal Images

Infrared(IR) microscope integrated the function of optical microscope and thermal imager. Due to the short depth of field of the IR microscope, it’s hard to obtain an IR image with all regions imaging clearly, especially when the surface fluctuation of the tested object exceeded the size of one depth of field. As a result,points which were not well-focused in IR image couldn’t reflect the temperature information accurately.This dissertation applied multi-focus image fusion to IR microscope to solve the problem mentioned above. The method got an all-focus image from multi-focus images. By collecting the IR images of the tested objects in different focal points of the same field of view and then fusing the multi-focus images, an all-focus IR image with all regions imaging clearly could be achieved. Then, by referring to the calibrated table which collected one-to-one relationship between gray value and temperature value, precise temperature values could be obtained. On the basis of multi-focus IR images sequence, 3D surface height of the tested objects was reconstructed, i.e. reconstructed the 3D surface height map of the tested object from multi-focus IR images sequence which contained depth information by depth from focus method. The major studies were summarized as follows:(1) A set of multi-focus images of working chip and thumb was captured by IR microscope. The thermal charge coupled device(CCD) translated from the top(bottom) to the bottom(top), and it was the same to the focus position moving direction. In Matlab, multi-focus image fusion was used to get an all-focus thermal image from the multi-focus images. The algorithm of image fusion was based on wavelet transform. Effect of different focus position intervals on the fusion results was analyzed. The error between temperature measurement based on multi-focus thermal images and that of single thermal image was calculated. It was concluded that the method based on multi-focus thermal images improved the temperature measurement error to at least 0.62oC(average sharing to every pixel of thermal image)in working chip and at least 0.10oC in the skin of thumb.(2) Based on the multi-focus IR images sequence which contained the depth information, chip’s 3D surface height map was reconstructed by the method of depth from focus in Matlab. Effect of different pre-processing, reconstruction operator, and post-processing on the reconstructed height maps was analyzed with surface roughness as the evaluation criteria. By comparing the reconstructed data with the standard data, the reconstruction quality was analyzed. In order to integrate the 3D surface height information with the temperature information, open source code(Open GL) was used to render the all-focus thermal image to the 3D reconstructed surface, forming 3D surface height map whose surface information was IR thermal image information.(3) Based the hardware platform of IR microscope, the user operational software was developed in Visual Studio 2012. With hardware controlled by the software,temperature measurement became more automated with z axis moving and image capturing simultaneously processing, which could meet the need for real-time capturing of multi-focus images.