Look Direction Error and Its Correction on Airborne Imaging Scanner Image

In aerial remote sensing, because the pixels of scanning line have different look directions they have different radiometric errors, called as the look direction error in this paper. The specular reflection and hot point are the examples of the look direction error. The radiometric errors affect the precision of digital analyses' result based on remote sensing images. So it is very necessary to correct the radiometric errors. The paper,beginning from both the sensor's look direction and the micro-plane reflection,analyzed and researched the look direction error to Airborne Imaging Scanner image and the corresponding correction method. The authors consider that in the airborne imaging scanner, the factors,mainly including micro-terrain and the change of relative geographic position in imaging,are the main reasons to create the look direction error. The radiometric error is not only related to the look direction,but also closely related to the spectral bands of sensors. Because of the complexity of the terrain distribution,it is very difficult to correct exactly the radiometric error. The authors also consider that the look direction error has the systematic and random features. According to the statistic theory,a correct method based on image's statistic analysis was put out and used to calibrate the look direction error. The main ideas of the correcting method include:the ratio between average reflecting energy and the reflecting energy received by sensor is fixed for every pixels in the same look direction;there are the same average reflecting energy distribution in the different look direction. The paper sees the earth surface as a model between full reflecting model and full Lambian model. The model is different from both full reflecting model and full Lambian model. In the paper's model,if the reflection intensity of every direction of object is the same the model becomes full Lambian model. If there is only one reflection direction to an input ray, the model is similar to full reflecting model. The determination of correcting coefficients depends on the distribution of reflection energy in the every look directions. At last,the experiment was completed by using simulated image data.