Multi-band Imaging Sensor Calibration

The traditional visible light cameras only record the components of the red,green and blue channels during the imaging process,which are the same as human eyes,but the three channels cannot describe the world precisely.So,those cameras are susceptible to ambient light,in material identification,safety monitoring,medical diagnosis and other fields,the effect is far from satisfactory.Non-visible light imaging devices and spectral imaging devices can record spectral information other than RGB and make up for the shortcomings of ordinary visible light camera in many areas.During imaging processing,light travels through the lens and projects onto the imaging chips,and the depth information is lost due to projection transformation.So positioning or other diagnosis work via single camera need special calibration work to be carried out in advance.What's more,military safety,environmental monitoring and other works employing multi-spectral imaging devices also require accurate calibration.This paper focuses on how to calibrate non-visible light band imaging sensors or hyperspectral capture devices more accurately.The main work and innovation points include the following aspects:1.Propose a method of calibrating the solar blind ultraviolet imaging devices.Compared with the conventional method,it can overcome the problem caused by the low resolution and lack of texture information of solar blind ultraviolet imaging device.The proposed method can match image points and object points in real time precisely,and the method can still work as usual even some reference points are lost which makes the system more robust.We achieve the goal of high-precision positioning via solar blind UV devices.2.Propose a method of locating the reflective medical ball via binocular infrared imaging devices.Compared with the conventional method,this method can get the imaging and spatial coordinates of the imaging ball more accurately,which makes the positioning more accurate.Using this method,we realize the goal of high precision fully automated medical positioning.3.Propose a method of calibrating and aligning the Prism-Mask Multispectral Video Imaging System that can align the two-cameras accurately and improve the spectral resolution and stability.Compared to the conventional method,the accuracy of the spectral data improves significantly.