The Compensation Of Color Remote Sensing Image

The needing of making color map in military, Tri-Linear color CCD remote sensing camera was researched. But, when the color remote sensing images were captured, they were affected by many factors. So, these images would be non-uniformity, geometric distortion, and color distortion, which must be processed in order to correct the effect of non-uniformity, geometric distortion, and color distortion before they would been further analyzed. In the aspect of non-uniformity correction, the linearity of photo electricity response of the remote sensing camera was calibrated. Just because that the result of remote sensing camera' s response is similar linear in the range of working, the images of Tri-Linear color CCD are corrected by two points calibration method. Then we get satisfactory results, the non-uniformity of image after non-uniformity correction is lower five times than it before non-uniformity correction, all the pixels' three values of the images are uniform.In the aspect of geometric correction, geometric correction of image captured by Tri-Linear color CCD remote sensing camera by colinearity equation is putted forward. Both constant geometric distortion induced by the characteristic of Tri-Linear color CCD and stochastic geometric distortion induced by the position of sensor and changing of pose are corrected. First, imaging equation of the remote sensing camera is set up, then geometric distortion model and geometric correction model are set up. Simulate experiment indicate that geometric correction of color remote sensing image by co-linearity equation can compensate well. In addition, this method can be finished on computer automatically, avoid error induced by people.In the aspect of color processing, that Tri-Linear color CCD remote sensing camera is putted into the system of color management is putted forward, namely, the color space of the remote sensing camera is transformed to the sRGB color space in order that the remote sensing images can been outputted on more output devices. At the same time, it would stabilize remote sensinginformation and benefit remote sensing image analysis, color object recognition and the combination of image captured in different time or by different remote sensing camera.Based on the characteristic of the sRGB color space, the whole transform processing is separated into two steps. First, transform from the color space of remote sensing camera to linear sRGB color space, then, transform from linear sRGB color space to non-linear sRGB color space. The second step is fixed, and the first step is more complex, because remote sensing camera doesn' t capture color patches, the color space transformation based on remote sensing camera response model is putted forward. The simulate experiment indicate that the color space transformation based on remote sensing response model can satisfy requirement, its excellence is flexible, and the color space transformation matrix can change with the changing of the parameter of sun and atmosphere.