Research On Target Detection And Recognition Method Based On Underwater Image Enhancement

The ocean is rich in resources,which play a decisive role in the continuation of life on earth.During the propagation of visible light underwater,due to the scattering of particles and the absorption of water,the obtained underwater image is difficult to satisfy human visual perception.The scattering of light easily causes the underwater image to become blurred and the details are blurred,and the absorption of light easily causes the color distortion of the underwater image,the reduction of the image contrast and the brightness of the image.Due to the absorption and scattering of underwater light,the detection and recognition of underwater targets face huge challenges.In recent years,underwater target detection and recognition has been widely used in marine fishery,marine military,marine protection,underwater archaeology and other fields.In the field of marine fisheries,real-time monitoring and tracking of fish schools can be carried out,and seafood can be detected and automatically captured;in the field of marine military,monitoring and tracking of underwater suspicious targets;in the field of underwater environmental protection,Can grab the seabed garbage and protect the environment;in the field of underwater archaeology,it can help archaeologists explore seabed sunken ships and underwater relics.Due to the intelligent and efficient application of underwater optical images,more and more researchers and scholars are engaged in the research of underwater optical images.The main research contents of this article are summarized as follows:(1)The enhanced restoration method of underwater image is summarized in detail.First,the physical imaging model of underwater optical images is introduced.Then,it specifically introduces the underwater optical image enhancement restoration algorithm based on non-physical model and physical model.Finally,this article evaluates the advantages and disadvantages of underwater optical image enhancement and restoration algorithms.(2)This paper proposes an underwater image enhancement algorithm based on fusion.In order to avoid the influence of noise on the experimental results,Gaussian filtering is used to process the acquired underwater optical images to remove image noise.Then,color correction processing(UCM)and contrast enhancement processing(CLAHE)are performed on this image respectively,and then weighted fusion processing is performed on the two images.Finally,compare the method in this article with other methods experimentally,and the experimental results are analyzed and evaluated subjectively and objectively.It is found that the image processed by the algorithm in this paper can reach the level of image chroma,image contrast,and saturation.Balance,so that the processed image is more in line with human visual perception.(3)This paper proposes a target detection and recognition algorithm based on image enhancement.First,this article introduces the basic network structure of convolutional neural networks in detail.Then,the YOLOv3 target detection network is analyzed in detail,and on this basis,a target detection and recognition method based on image enhancement is proposed,and the data set,underwater image preprocessing method and underwater image target detection method evaluation index are introduced..Finally,through experimental comparison,compared with the YOLOv3-DPFIN,YOLOv3 and YOLOv3-Tiny algorithms,the enhanced target detection and recognition algorithm in this paper is superior.