Airborne 3D imager which integrates global positioning system

attitude measurement unit

scanning laser range finder and spectral imager can produce geo-referenced image and digital terrain models （DTM） without ground control points. The raw image is rectified based on the laser points with known coordinates. Owning to the existence of the error

the geo-referenced images of neighbor strips can not be mosaiced seamlessly. Therefore

it is necessary to eliminate the error. The error source of each component of 3D imager such as GPS error

attitude error and laser range error is analyzed and the system error between the components is also described. Then we focus on the error processing algorithm. The system error is determined and corrected using least mean gray difference in the overlap and the random error is processed by variant weight algorithm in order to mosaic the strips. The test result proves that the error processing methods is feasible.