Abstract: Image point offset is one of the main factors affecting the accuracy of bundle block adjustment. For a long time, how to overcome the influence of the offset in adjustment, and then improve the accuracy of the block adjustment, has received close attention and great importance from scholars in this industry. At present, the conventional method to overcome image point offset is to describe it with a specific parameter model, build an additional parameter model-based self-calibration bundle block adjustment and solve the introduced parameters while doing adjustment to realize the compensation of systematic error. Among the existing algorithms, Bauer model, Ebner model, Brown model and optical distortion model are common models to describe image point offset. In order to verify the effect on systematic error of the above four models in block adjustment, this paper designs self-calibration bundle block adjustment models corresponding to the four models based on classic bundle block adjustment, and implements them by programming. Finally, the algorithm results are verified, compared and analyzed through examples. Experiments show that introduction of Ebner model and Brown model can effectively improve the accuracy of bundle block adjustment; introduction of Bauer model is more effective in improving the accuracy of small block adjustment; introduction of optical distortion has a relatively little effect on improving block adjustment accuracy.

Key words: image point offset, aerial triangulation, block adjustment, systematic error, bundle