Research On The Method Of Long Distance And Fast Recognition Of UAV

With the rapid development of UAV industry and image recognition technology,it is more and more common for UAV to collect target image information in cruise mode.Image recognition technology is changing with each passing day.The recognition of long-distance target image by UAV has become the future development trend.Using computer image recognition technology and taking UAV airborne equipment as cloud transmission carrier,it is a new requirement for UAV to automatically and quickly analyze and recognize the fuzzy target image.In order to enhance the ability of the target image to be recognized and recover the original target image,this paper proposes a recognition strategy which can determine the ability of the original target image's recognize in the process of contouring information about the target image,and makes a deep research on the related image recognition and processing technology.Aiming at the fuzzy image preprocessing problem of the target image which is acquired by UAV,this paper puts forward the image fuzzy scheme based on Lucy-Richardson restoration operator and determines to act it.At the same time,this paper uses the superposition Wavelet Bi-cubic matching algorithm which aims to process the fuzzy image to solve the problem of image distortion.In the process of image cloud transmission,the UAV is realized by combining electronic anti shake technology and image compression technology fast image transmission.In view of the imprecise problem of image recognition caused by the fast speed and long target distance during the low altitude flight of UAV,this paper proposes a method of contour matching recognition based on Bessel fitting,and proposes an image contour matching algorithm based on Bessel fitting and Hu invariant moment which is on the basis of optimizing this algorithm.Combining with seven Hu invariant moment parameters representing contour similarity,the target is to be recognized different images and original images which are matched by different criteria and verification methods.In this paper,according to the phase operations of Bessel fitting contour,edge extraction,feature point sampling,feature matrix andsimilarity measurement used in the algorithm process,the research of image matching and recognition is carried out with the static standing human object.