Using images acquired by medium format CCD sensors from aerial platforms is an alternative to reduce costs of aerial surveying projects. However, using sensors that capture only the visible band (RGB) it restricts some applications, making the integration of an additional sensor that captures the near infrared (IR) important. There are several commercial solutions for the simultaneous acquisition of multiple spectral bands. A cost effective alternative is the integration of a RGB camera to a second one, to capture the infrared band. It is necessary, then, to register the images, which involve the determination of a set of corresponding points or features, to calculate a polynomial mapping function and to resample one of the images. One of the most critical problems is the determination of corresponding points between the images due to differences in the images radiometry. Some techniques for locating interest points and a correlation function using the differences in magnitudes and directions of the gray levels gradients RGB and IR images, are proposed in this paper. Several experiments using this technique were conducted, indicating that it is possible to use it, achieving residual errors less than 1.5 pixel.
Correspondence between RGB and IR Images; Image Registration; RGB and IR Registration
