Design And Implementation Of Sunken Oil Recognition Software Based On Acoustic Image

With the continuous exploitation of global offshore oil resources,offshore oil spills have increased.Once oil spill at sea occurs,it will cause harm to the ocean environment for several years or even decades,so it is heavily concerned by countries today.After oil spill disastar,how to quickly and effectively monitor the scope of oil spills is the top priority for effective oil spill management.In this paper,research is carried out based on sonar images obtained from two types of acoustic detection devices,forward-looking sonar and side-scan sonar.The target recognition of sunken oil is completed through the combination of sonar images and pattern recognition methods.Design and development of software for the target recognition method of the primer are completed.Firstly,the imaging principles of side-scan sonar and forward-looking sonar are described in this paper,image distortion,denoising and image enhancement are pre-processed.Second,the pre-processed sonar image is combined with image segmentation,target feature extraction,classifier training,and target recognition using the trained classifier for target recognition in the sonar image.Among them,the Markov image segmentation method,the gray statistical characteristics of the image and the statistical feature extraction based on the gray level co-occurrence matrix are studied.Finally,the support vector machine-based classification and recognition method is used to identify the submerged bottom oil target.Third,take Py Charm2018(PyQt)as the development platform and the sonar image-based target recognition technology,the sunken oil target recognition software was designed as image aid preprocessing,image denoising,and image Segmentation and submersible oil target recognition four modules complete the design and development of submersible oil recognition software based on forward-looking sonar images and side-scan sonar images.The main functions of this software are: reading,displaying and saving functions of common sonar raw data Extended Triton Format(*.XTF),Blueview Sonar Data(*.son)format files;speed correction,grayscale of side scan sonar images Correction and image enhancement of sonar images;common spatial domain filtering and denoising functions based on sonar images;wavelet denoising;maximum inter-class variance method,C-means clustering segmentation method;and Markov random field model-based Image segmentation function;feature extraction function based on image gray level statistical method and gray level co-occurrence matrix;support vector machine classification method based on sunken oil target recognition function.During the functional verification of this set of sonar image-based software for sunken,functional verification was performed using forward-looking sonar data and side-scan sonar data.The analysis of the results obtained by processing the experimental data proves that the software functions and The algorithm effect can meet the expected requirements of software design,and at the same time,it can stably complete the target recognition task of sunken oil.