新型无人机图像序列压缩与重构算法

New Image Sequence Compression and Reconstruction Algorithm for UAV

针对无人机图像序列帧间相关性强的特点,提出了一种基于压缩感知的无人机图像序列压缩与重构算法。在编码端生成每幅图像的随机测量值和飞行参数,通过数据链路传输至地面站解码端。在解码端,通过分析摄像机与物体间的几何关系建立了运动估计模型,从而减少了图像间的冗余。去相关后的图像更稀疏,重构也更容易,并且重构后图像具有更高质量。试验结果表明,该算法不仅可以提高重构图像的峰值信噪比(PSNR),且有效降低了编码端的工作时间,具有较好的实时性。该算法计算复杂度低,硬件实现较简单,适用于无人机系统。

Aimed at the characteristic of the strong inter-frame correlation of unmanned aerial vehicle(UAV)image sequences,a compression and reconstruction algorithm for UAV image sequences based on compressed sensing is proposed.Random measurements and flight parameters of each image are generated at the encoder,and are transmitted over the data link to the ground station decoder.At the decoder,by analyzing the geometric relationship between the camera and the object,a motion estimation model is established.Thus,the redundancies between images are reduced.The de-correlated image is sparser,the reconstruction is easier,and the reconstructed image has higher quality.The experimental results show that the algorithmcan improve the peak signal to noise ratio(PSNR)of the reconstructed image,and theworking time of the encoder is effectively reduced.The algorithmhas better real-time performance.Due to the low computational complexity,the hardware implementation of the algorithm is simpler and suitable for UAV systems.