复杂海面环境下船只边缘识别算法的改进  

Improvement of Edge Recognition Algorithm for Ships in Complex Sea Surface Environment

作　　者：童强 李太君[1] Tong Qiang;LI Taijun(College of Information Science and Technology,Hainan University,Haikou 570228,China)

机构地区：[1]海南大学信息科学技术学院,海南海口570228

出　　处：《海南大学学报（自然科学版）》2018年第3期235-240,共6页Natural Science Journal of Hainan University

摘　　要：针对复杂海面环境下船只边缘识别问题,提出了一种融合Retinex图像去雾预处理和改进自适应阈值SUSAN边缘检测的算法,用来提取船只目标边缘特征.对由于不同场景海雾造成的图像模糊,采用Retinex预处理增强图像信息,然后对待检测目标像素剔除伪边缘点,采用自适应算法获取几何阈值t和最大类间方差(Otsu)的方法选取双阈值g,进而对图像进行边缘点检测、提取边缘特征.实验结果表明,融合后的算法能够有效提高复杂海面船只边缘检测的鲁棒性.In the report,aiming at the problem of vessel edge recognition in complex sea surface environment,an algorithm based on fusion Retinex image defogging preprocessing and improved adaptive threshold SUSAN edge detection was proposed to extract ship edge features. For image blur caused by sea fog in different scenes,Retinex preprocessing was used to enhance the image information. Then,the target pixel was removed from the false edge point,and the adaptive algorithm was used to obtain the geometric threshold t and the maximum betweenclass variance（ Otsu） g. Lastly,edge point detection of the image was performed and edge features were extracted. The results showed that the fusion algorithm can effectively improve the robustness of edge detection of complex sea surface vessels.

关 键 词：SUSAN算法 去雾算法 边缘检测 自适应几何阈值 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]