Information Extraction And Utilization Technology Based On YellowFin Sonar

With the continuous development of sidescan sonar system,side scan sonar is widely applied to various fields,but its fast accuration on underwater target recognition is based on the extraction of information utilization.This article is based on YellowFin side scan sonar detection and the identification of certain goals through analyzing the characteristics of the side scan sonar image,adopt the method of artificial and cooperate on the specific target feature extraction and classification recognition.In this paper,we first introduces the working principle of side scan sonar and image characteristics,and analyzed the function and performance index of YellowFin sonar.On the basis of the data information extraction,we adopted the convolution image recognition method of neural network to study the unmanned automatic identification of crude extraction technology.this part refer to the LeNet-5 recognition model of network structure and implement the no identification of specific target.In the image enhancement process of the extracted potential target image,the extracted target image is enhanced and optimized through the typical de-noising and segmentation algorithm.we compared the advantages and disadvantages of conventional median filter,adaptive median filter and wavelet denoising and studied the thresholds iterative side-scan sonar image segmentation algorithm and compared it with the histogram threshold method and FCM method on the image segmentation results.In the feature extraction and recognition part of the side scan sonar image,the algorithm of edge detection is introduced to extract the image features.The feature of image is extracted by the method of edge detection and mainly introduces the Sobel operator,the Laplasse operator and the Canny operator.In the image recognition part,the SIFT feature matching recognition algorithm is introduced and the simulation experiment on the stability and generality of the algorithm is carried out.In view of the convolution network access screen side scan sonar image,we are using filtering enhancement,image segmentation,edge extraction and improved accuracy of the side scan sonar image recognition after further feature matching recognition.In this paper,according to the requirements of YellowFin sidescan sonar searching for specific targets on unmanned platforms,we propose to use the convolutional neural network to perform coarse screening on side scan sonar image targets based on the training of known target sample sets to target suspected targets.Quickly obtain,then use post-processing to further enhance the information processing process of recognition performance,and carried out simulated target test and performance test.The processing results verify the feasibility of the idea.