Fast Processing of UAV Remote Sensing Image Based on DSP Hardware Simulation Platform

Abstract

The traditional unmanned aerial vehicle (UAV) remote sensing image processing algorithm is difficult to guarantee both accuracy and speed when dealing with massive data of UAV network. This paper designs an on-the-spot processing system based on portable digital signal processor (DSP) hardware simulation platform. The bilinear interpolation algorithm is optimized and improved for the adaptability of the DSP platform with the reference of the bicubic interpolation algorithm, which is long time-consuming but high accuracy. The nearest neighbor interpolation is merged in the bilinear interpolation algorithm, which improves the edge blurring problem of bilinear interpolation algorithm. In the loop calculation of the multiplication and addition iteration calculation segment of the algorithm, the parallelism of the instruction level is increased by increasing the instruction level parallelism between the loop bodies. Experiments show that, on the basis of ensuring the accuracy of the bilinear interpolation algorithm, the execution speed of the processing algorithm is improved, which can meet the requirements of fast real-time processing of remote sensing images of complex unmanned aerial vehicles.