基于海天线检测的无人艇晃动条件下视频拼接方法

Video stitching method for unmanned surface vehicle under sloshing condition based on sea-sky detection

为获得无人艇周围的大视场无失真图像信息,需要对无人艇周边图像进行实时拼接处理。该文采用单应性映射方法对图像进行拼接。无人艇在行驶过程中不可避免的会发生一定程度的晃动,通过建立空间数学模型,得到晃动条件下的图像拼接单应性矩阵计算方法。并通过对图像中海天线的检测对船体晃动进行空间分解,计算。海天线的检测采用改进的概率霍夫方法。采用该文方法对海天线进行检测,正确率达到94.5%,且该方法对逆光图片具有鲁棒性。通过拼接3台相机的视频图像得到无人艇正前方180°视场。该方法实时性良好,满足实际需求。

In order to get the large field of view,undistorted image information around the unmanned surface vehicle,the surrounding images of the unmanned surface vehicle need to stitched.In this paper,homography mapping is applied to image stitching.A certain degree of sloshing will inevitably occur during the course of unmanned surface vehicle.By establishing a spatial mathematical model,a method for calculating the homography matrix of the image stitching under the condition of sloshing is obtained.The sloshing is decomposed and calculated through the detection of the sea-sky line in the image.An improved probability Hough line detection method is used for the detection of sea-sky line.By using this method to detect the seasky line,the correct rate reached94.5%,and this method is robust to the backlighting image.By stitching the video images of three cameras,can get the180degree view field in front of the unmanned surface vehicle.The actual test is carried out in the actual hardware environment of the unmanned surface vehicle,reveals good real time performance to meet the actual needs.