低空航拍视频中基于Vibe算法的船舶检测方法  

Ship detection method in low aerial video based on Vibe

作　　者：蒋晶晶[1] 安博文[2] 

机构地区：[1]上海海事大学信息工程学院电子工程系,上海201306 [2]上海海事大学信息工程学院,上海201306

出　　处：《微型机与应用》2017年第10期44-47,共4页Microcomputer & Its Applications

基　　金：国家自然科学基金(61171126)

摘　　要：针对低空航拍视频中的船舶检测研究普遍存在因背景变化及水纹扰动导致的识别效果不理想等问题,提出了一种基于改进型Vibe(Visual background extractor)算法的船舶检测方法。该方法在传统Vibe算法基础上融合了改进型的Canny算子,采取平滑滤波与自适应阈值分割策略提高了船舶轮廓提取的准确性。利用四方位结构元素的形态学机制,降低了水面波纹对目标识别的干扰影响。对于船舶检测中的内部空洞则选取了种子填充法进行填充,最后根据船舶外轮廓实现对船舶目标的框定。实验结果表明,所提出的改进型Vibe算法的船舶检测效果明显优于传统方法,验证了其可行性与实用性。Ship detection in low-altitude aerial video universally has not ideal performance due to background disturbances and ripples. In response to the issues,a new method is proposed based on improved Vibe( Visual background extractor). This saliency-mixture of improved Canny operator consists of smoothing and adaptive thresholding strategy to realize comparatively higher accuracy of ship contour extraction. Morphological mechanisms of the four-way structural elements reduce interference caused by ripples in target recognition. Internal cavity in ships model is filled though seed filling algorithm,finally,ship contours are identified precisely according to the outline of its characteristic. The experimental results demonstrate that the proposed method is significantly superior to the traditional method,and its feasibility and practicality have been proved.

关 键 词：船舶检测 Vibe算法 CANNY算子 种子填充法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]