Design And Implementation Of Underwater Pipeline Leak Detection Method Based On Forward-looking Sonar

The main purpose of this paper is through the study of underwater leakage gas detection methods based on image target recognition,provide basis for submarine pipeline leak hunting by underwater robot with forward looking sonar.In this paper,the research can be divided into three parts:By studying the procedure of image preprocessing,target feature extraction and classification,the median filter,threshold segmentation are chosen as the preprocessing methods;the connected component labeling is chosen as suspicious area extraction method;the shape feature extraction method based on invariant moments and texture feature extraction method based on gray level co-occurrence matrix are used;the classification method based on the perceptron is embodied.Formed a set of leak gas recognition algorithm with high identification correctness and real-time property.Embedded operating system VxWorks and PC/104 have the advantages of high real-time performance and excellent reliability.They were often conjunctively used as control system of underwater robot and general information processing system.So,choose to transplant the algorithm to the system composed of PC/104 and VxWorks.Design the module by using the multitasking programming methods.This module has the functions of receiving sonar data,image generate and processing,processing results back haul.Design display software on PC.The software has the core functions of simulating sonar system to send the sonar data to the embedded pipeline leak detection module,receiving processing results from it and displaying the results.In addition to the above core functions,the software also has functions of display enhancement,schedule control,target distance forecast,recording parameter display.Finally form detection system for leak gas composed of a PC/104 and a display software on PC.The workflow of the system is: PC display software sends sonar data to the embedded pipeline leak detection module,the module identifies whether the image contains leak gas and return the result to the software on PC to display.