全景海域图像中的圆形海天线提取

针对全景海域图像背景复杂且海天线呈近似圆形的特点,提出了一种基于分形维数和改进Hough圆变换的全景海天线提取算法。该算法首先通过地毯覆盖法计算海域全景图像的分形维数,并据此提取出全景采集装置等设备在全景图像中的成像,消除其对海天线检测的不良影响。针对经典的梯度Hough圆变换算法提取全景海天线时无法给出唯一正确解的问题,提出了一种改进的Hough圆变换算法来有效提取全景海天线。对400幅不同情况下采集的全景海域图像进行了实验。结果表明,提出的算法在理想情况、海天线部分缺失、低对比度和海天线大范围断裂等多种情况下均可以有效提取出全景圆形海天线,检测准确率达95.75%,适用性和鲁棒性良好。

Circle sea-sky-line extraction in panoramic images

For a panoramic sea image with complex backgrounds and the sea-sky-line shows as a nearly circle, this paper proposes a panoramic sea-sky-line extraction algorithm based on fractal dimensions and improved Hough circle transform. It firstly calculates the fractal dimensions of the panoramic sea image by blanket-covering method, then extracts the images of panoramic acquisition devices and other equipment to eliminate their adverse effects on sea-sky-line detection based on these fractal dimensions. As the traditional gradient Hough circle transform can not give out a correct solution when it is used to extract the panoramic sea-sky-line, an improved Hough circle transform algorithm is put forward. 400 frame panoramic sea images extracted in different conditions are processed. Experimental results show that this proposed algorithm effectively extracts the circle panoramic sea-sky-lines under conditions of ideal background, lower contrasts images and the sea-sky-line with partial loss or severe breaking and its detection accuracy reaches 95.75%. The applicability and robustness of the proposed algorithm are excellent.