Fine-grained parallel algorithm for remote sensing image mosaics for cluster system

The combining of image mosaics is an important part of remote sensing image processing. A fast fine grained parallel algorithm was developed to produce seamless image mosaics with global grayscale coherency. The algorithm uses a double buffer queue and a dynamic task selection strategy to greatly reduce the communication overhead and task waiting time. The equalization and resampling procedure can be easily embedded into the parallel model. The algorithm also includes an improved image merging algorithm which supports irregular boundary image merging. Experiments to test the efficiency and effectiveness of the algorithm show that the new algorithm has approximate linear speedup and provides very good visual results.