Research On The Rapid Detection Technology Of Vision-oriented UAV Aerial Targets

In recent years,the rapid development of UAV(Unmanned Aerial Vehicle)technology makes UAV ground target detection technology has become an important research direction in the field of computer vision.UAV has a wide range of applications in military investigation,traffic control and other scenarios.However,the UAV images have many problems such as low target resolution,scale changes,environmental changes,multi-target interference and complex background environment.In this paper,based on the above difficulties,the key technologies of moving target detection in UAV videos are deeply explored and a set of UAV flight management and evaluation system is designed to conduct scientific planning and evaluation.This paper has used the data enhancement method to expand the training samples and selected the best data enhancement scheme through experiments.This method can effectively avoid the over-fitting caused by the shortage of training samples.Focusing on the problem such as the degradation of quality and blurring of the aerial images in haze weather,this paper has selected the method of image haze removal based on the detection of haze line,located the haze line by clustering and estimated the transmission coefficient.Regularized the coefficients and finally got the haze removal image.Derived from the original SSD target detection algorithm,this paper has used a residual network with better characterization ability to replace the basic network and a residual learning to reduce the network training difficulty,improving the target detection accuracy.By introducing a hopping connection mechanism,the redundancy of the extracted features is reduced,and the problem of performance degradation after the increase of the number of layers is solved.The effectiveness of the algorithm is verified through experimental comparison.Aiming at the problem of target repeated detection and small sample missing detection of the original SSD target detection algorithm,this paper has proposed an aerial target detection algorithm based on feature information fusion.By integrating information with different feature layers,this algorithm has effectively made up for the difference between low-level visual features and high-level semantic features in neural networks.Results show that the algorithm has good performance in both detection accuracy and real-time performance.Finally,we summarize the full-text work and look forward to the future work.