SPOT Image Geometric Precise Correction With DEM Auxiliary

It is possible for us to observe the surface in small details, draw the large scale of remote sensing maps and monitor the influences on environment from human activities when high-resolution satellite imagery appears, and it may have broad application. But the information of the high resolution satellite imagery is multiplex, multiform and complicated, this will bring a series of problems to the application of the related field, and the first problem that we need to solve is the imagery geometric correction.The generalized model that don't rely on sensors tend to improve now, and the geometric correction precision has increased significantly.However, the geometric process of space imaging is not considered in the general geometric correction models, but analogy digitally is directed to deformation of image itself. If we regard the image approximately as the plain region, it is hard to ensure the precision of the adjusted image as an average elevation adopted to do geometrical transformation when the actual terrain is greatly undulating. So it is necessary to consider the digital elevation model (DEM) used to implement precision correction, direct to high and steep area especially.In this paper, we use SPOT high resolution image with a resolution of 10m and the aerial data with a resolution of 2m to correct. In order to strengthen the contrast, the SPOT image is cut as the plain region and the mountain area. We do geometric correction with no DEM auxiliary data based on quadratic polynomial model,cubic polynomial model, TIN grid model and rational function model, and make a comparative analysis for the precision of geometric correction. At the same time, we make a comparative analysis with DEM auxiliary data in the plain region and mountain area, and get some conclusion as follows:1. The correction accuracy based on cubic polynomial model is the same as that based on TIN grid model in the lateral correlation, and the rational function model is the best of all models. 2. The correction controls'accuracy in mountain area is lower than that in the plain region obviously due to the complexity of the terrain. The correction accuracy with no DEM auxiliary data in mountain area is lower than that in the plain region too.3. There has been a significant improvement in accuracy no matter which model of the four was used with DEM auxiliary data in the plain region and mountain area, and the correction accuracy's improvement is more obvious based on the cubic polynomial model and TIN grid model.4. The correction accuracy with DEM auxiliary data in the plain region and mountain area are both have a significant improvement taking into account of the actual topography, up to 80and 88percent separately ,and the correction accuracy's improvement in mountain area is more obvious than that in the plain region. This show that, assign a mean elevation to the ground points, but ignore the ups and downs of the terrain, the correction accuracy is not meet the requirements.