摘要：Panoramic image pre-processing is one of key technologies in subsequent image processing and applications. An adaptive pre-processing multi-layer model for the sequences of panoramic images formed by six cameras is proposed in this paper. The algorithm develops a luminance classification criterion for different level of panoramic image based on the distribution of the statistical characteristics and threshold judgment. The criterion is able to cope with the image pre-processing effectively and stably when luminance altered. By this method, each pixel is adaptive based on illumination levels, resulting in the development of visual effect. Experiment results show that the proposed algorithm can significantly improve the quality of 3D panoramic images in complex environments.
