Researches On DIDSON-based Fish Quantitative Assessment Technology

The Dishui Lake of Shanghai Lingang Town was completed to its excavation, andwater diversion. The area of Water lake is5.56square kilometers, the average depth is3.7m, and the most deep place is6.2meters. Since the Dishui Lake was completed,the management of the lake and river channel spent about315ton fry from2007, breedis mainly silver carp. After years of launch, at present the lakes fishery resources need toinvestigate, it will provide necessary data for the further development andmanagement.In this paper, we use the DIDSON for Dishui Lake on the fishery resources survey.According to the experiments identification sonar in the investigation of data, and thecharacteristics of the data, we convert the showing of the right angle coordinate systemimages to the sonar image data storage polar coordinates, we use linear interpolationmethod to replenish image data, and then apply Median filter method to deal with thenoise for the noise of the data on the smooth. After treatment of data on the fish to beautomatic tracking count, the application background removal and open computingpixels are used to segment and extract for division of the data. Using the algorithm toextract skeleton of the length fish extract estimates, and using kalman filter algorithmto fish in the data tracked count. After tracking count the data, using statistical methodto the whole length of the fish on the analysis of the distribution. According to the GPSlocation to draw the fish district of distribution in the lake, We use KrigingInterpolation to water depth data calculation, and combined with distribution of fishand water depth distribution and analyzes the relationship, use the average densityrespectively method and partition method to estimate the total amount lake density.The results were as follows:(1) Due to the DIDSON from acoustic lens more far place, the data collected aboutthe intervals more wider (or so the lower resolution) to make about excessive spacingbetween data for processing the data and displaying to DIDSON. We use the linear interpolation method to deal with the image data interpolation to make the image datapointed in space become smaller, and looked more natural and clear. The data ofinterpolation using median filtering de-noising smooth processing to except for somenoise, very good to go, and make the image becoming smooth. Using vc++6.0,OpenGL and GLUT the image processing, display, such as inputting operationmanagement tasks.(2) To count the main processing and fish tracking of dealing with the image data,there have three steps: Fish target detection and division, fish body target analysis, fishtarget tracking. We use background removal, open computing pixels removal andbinary method to combine the fish target detection and division. And we calculatedivision of the center of the body after fish position and through the skeleton of theextraction method is used to calculate the length of the fish. In this paper, the kalmanfiltering algorithm is put use to fish target tracking. In a full time series of data12frames track processing found that, among of which only five consecutive frames thefish is tracking data, analyze the causes, and is the main front six frames the fish ofdata body target strength and similar backgrounds, it cannot extract the target fish fromthe background, because the fish target strength is low as noise removed form the lastof frames.From the track of the continuous5frames data shows, the use of kalmanfiltering algorithm can be very good for tracking fish, fish trajectory was cleardescription.(3) According to the statistical data tracking, we draw out the distribution chart offish in the lake, and combined with the water depth data, it is found that the fish ismainly concentrated in deep water area. In the middle and upper waters, there is almostno detection fish. According to the actual detecting data and using the average densitymethod, we found the number of fish is338193tails, about962835kg. By usingzoning density method, we found the number of fish is345253tails, about939885kg.