提出基于机器视觉技术的船舶航行危险区域自动识别方法，最大程度规避船舶航行风险。利用机器视觉技术获取船舶航行图像数据，并结合像素平滑滤波和帧间差分法去除原始船舶航行图像所含噪声。采用二阶高斯-马尔科夫机场算法对去噪后船舶航行图像的显著性区域候选节点作信息弥散处理，获取船舶航行图像显著图，通过均值偏移算法处理船舶航行显著图像的特征空间，获得多个分割区域后，在显著图中求解各区域的显著性均值，通过与阈值作比较，实现船舶航行危险区域识别。实验结果表明：该方法可有效提升船舶航行图像的视觉效果；生成的显著图细节完整；可实现不同危险区域的识别，识别效果突出。
An automatic identification method of ship navigation danger area based on machine vision technology is proposed to avoid the ship navigation risk to the greatest extent. Use machine vision technology to obtain ship navigation image data, combine pixel smoothing filtering with inter frame difference method to remove the noise contained in the original ship navigation image, use second-order Gauss Markov airport algorithm to diffuse the information of candidate nodes in the significant area of the denoised ship navigation image, obtain the prominent image of the ship navigation image, and use the mean shift algorithm to process the feature space of the prominent image of the ship navigation, After multiple segmented regions are obtained, the significance mean value of each region is calculated in the saliency map. By comparing with the threshold value, the ship navigation danger area is recognized. The experimental results show that this method can effectively improve the visual effect of ship navigation images, the details of the saliency map generated are complete, it can realize the identification of different hazardous areas with outstanding recognition effect.
2023,45(3): 157-160 收稿日期：2022-09-09
DOI：10.3404/j.issn.1672-7649.2023.03.030
分类号：U675.1
基金项目：广西壮族自治区广西高校中青年教师（科研）基础能力提升项目（2020KY30010）
作者简介：段仕浩(1982-),男,副教授,研究方向为大数据技术、物联网及人工智能
