Research On Underwater Cable Detection Method Based On RGB Channel Fusion

As people’s demand for resources continues to increase and resources are continuously developed and utilized,the development and utilization of the ocean has gradually become one of the research hotspots.For underwater positioning such as underwater cable detection and pipeline inspection,underwater unmanned aerial vehicles are usually used as carriers,and underwater vision systems based on cameras are used for underwater cable detection.For the traditional underwater cable detection method,the conventional grayscale algorithm and the filtering algorithm are usually used to preprocess the underwater color image.Due to the absorption of water and the effects of scattering,the grayscale effect of the underwater grayscale image and the non-target noise spot appear,resulting in poor target contrast.When using the traditional Otsu algorithm for image segmentation,it is easy to cause the underwater target and background segmentation boundaries to be inconspicuous and connected.Therefore,this paper proposes a grayscale method based on RGB channel fusion.This method firstly performs preliminary segmentation of the image on the color image for the approximate location of the target position of the image,and calculates the local window by creating it.The target local contrast in the window is obtained by re-melting the three color channels to obtain the final grayscale enhanced image,and verifying its practicability by serializing the image experiment,the method solves the underwater camera image The underwater target and background contrast are constantly changing and it is impossible to dynamically estimate the problem of RGB channel weight processing.By comparing the image and data processed by the method,it is found that the method greatly improves the local contrast and reduces the influence of image noise.By studying the traditional Hough line detection method,this paper finds that although it can detect the edge line of underwater cable,it is easy to be affected by image noise to form a false line.This paper uses the SVM and MLP two classifiers to classify the acquired binary images into the edge of the underwater cable image.Then,through the improvement of the algorithm of Hough,by modifying the voting formula of the polar coordinate system,the edge detection of the underwater cable will avoid false lines due to irregularities such as noise,shadows and similar lines.It is proved by experiments that the method adopted in this paper avoids the generation of false lines very well.The improved Hough method is used for edge detection,and the position of the two center lines is calculated as the position of the underwater cable.Finally,the imaging model of underwater camera is analyzed and the imaging geometry model of underwater cable detection is constructed and constructed.According to the established imaging model of UUV underwater cable,the conversion relationship between each coordinate system is derived.And the pixel coordinates of the underwater cable image are converted between the respective coordinate systems to obtain the position information in the UUV coordinate system,and the final underwater cable is estimated by the absolute position of the UUV latitude and longitude relative to the underwater cable image.The latitude and longitude,and the experiment to record the data,to estimate the final coordinate position,proves the effectiveness of the method.