Rectification Algorithm for Linear Pushbroom Image of UAV

SUMMARY This research presents a method of rectification for the image flow acquired by UAV. The sensor on the UAV is a multi-channel push broom sensor. The linear array can cover a stripe with approximately 6 Km on the ground. The motions (roll, pitch, yaw, etc) of the UVA platform are influenced by the relatively low-altitude flying so that have the attitude changed which are terrible to produce severe distortion of image. Therefore, a three-step rectification method is developed by means of the DEM and GCPs (ground control points) of study area, as well as data of the positioning instruments (GPS, INS) on the UVA platform. Firstly, The external orientation parameters (position, attitude, six for each image line) of each linear array are calculated with data provided by the positioning instruments (GPS, INS), and then, the collinearity equation for multi-lens sensors is adopted with combining with the DEM data, therefore, the original image can be rectified to level one (L1) image, which meets the requirements for accepted visibility. Secondly, a reference image is used and several GCPs are selected in order to produce Level two (L2) image with enhanced polynomials. The third step of the rectification is an essential process using partial distortion model which geometrically stretch the local image. Many experiments have been carried out in different conditions and areas using the method mentioned above. The results present the three step rectification is feasible and effective.