| Unmanned Aerial Vehicles(UAV),as a flexible high-speed air vehicle,can accomplish various difficult tasks by carrying auxiliary equipment.Aerial photography is a basic function.At the same time,the limitations of human beings and the demand for real-time information make the application of real-time aerial drones and related technologies continue to develop.The video data collected is generally large.Limited by the wireless network transmission bandwidth,video data needs to be compressed and processed in real time in order to transmit back to the ground in real time.However,the resolution of video is getting higher and higher,such as 4K,8K,etc.Processing video data is a highly complicated process which means high power consumption.It is an urgent problem to improve the energy efficiency of the UAV drone video acquisition,transmission,and playback with the drone power is limited.Based on the above objectives,this paper studies the three parts including the energy efficiency optimization,video compression and image reconstruction of UAV aerial video.The main research contents are as follows:1.Analyzing the energy consumption composition and influencing factors in the drone video acquisition,transmission,and playback process,this paper demonstrate that there is a close relationship among drone energy consumption flight speed,hardware equipment,resolution,and working time.The DJ drone aerial photography system was used for verification experiments.The experimental data proves that the video with the same resolution can consume the same average amount of electricity per unit of time,and the higher the resolution requirement,the more electricity is consumed per unit of time.2.In this paper,combining the characteristics of UAV aerial motion,which is similar to the straight line motion in a single direction,a compression method is proposed.that is,a compression method based on affine model which uses a six-parameter affine model to solve the problem.The global motion of the human-machine video is described and the video is compressed according to the global motion parameters.The experimental results show that compared with the compression algorithm H.264,this algorithm reduces the compression time largely on the premise of guaranteeing the compression performance,and it can meet the quality requirements of the video transmission of the drone and improve the energy efficiency of the compression part.3.For the campus patrols and other fixed areas,with the super-resolution enhancement and the grayscale image colorization technology merged,the Cs RGAN(Color-super-Resolution Generative Adversarial Network)model is proposed.To reduce drone video data energy consumption,grayscale image resolution at the sky end compressed and transmission can be reconstructed to color high-definition video images in the ground end.The results show that while ensuring the quality of the video,less data is collected and the UAV power utilization is maximized,which proves the feasibility of this method.4.The evaluation indicators adopted in this article include not only the battery evaluation standard MPB(Minutes Per Battery),but also a more comprehensive consideration of the solution from multiple dimensions in terms of video quality,resolution of video capture and playback,and transmission time.For the fixed area,the effects of the Cs RGAN model and the affine model-based compression algorithm can doubtlessly improve the energy efficiency of the UAV,and the former is much better than the traditional H.264 and affine model-based compression. |
PDF Full Text Request