摘要
针对可见光图像中的舰船目标检测与识别问题,提出了一种基于图像显著性特征(Visual Saliency Features,VSF)的舰船目标检测识别方法。该方法包括3个主要内容:计算图像各通道全局对比度特征值,融合并进行分块压缩后得到显著图;采用Otsu算法自适应分割图像,获取舰船目标待识别区域;使用HOG特征+SVM分类器对候选区域图像进行验证识别,得到舰船目标检测识别结果。实验结果表明,提出的算法能够有效检测识别舰船目标,检测识别准确率高,算法鲁棒性好。
In order to solve the problem of ship detection and recognition in visual images,an algorithm based on VSF(Visual Saliency Feature)is proposed.The method consists of three main contents.First,it uses saliency detection algorithm to get the global contrast maps of three channels,then merges each channel and compasses to get the final saliency map.Second,split images use Otsu algorithm to obtain the regions of ship targets.Third,it uses HOG feature+SVM classifier to verify and recognize the image of the candidate area,and obtain the result of ship target detection and recognition.Experimental results indicate that the proposed algorithm can effectively detect and recognize ship targets,with high detection and recognition accuracy and good robustness.
作者
李宗鑫
王启曙
于娟娟
LI Zongxin;WANG Qishu;YU Juanjuan(Unit 92001,Qingdao Shandong 266000,China;Yantai Preschool Education College,Yantai Shandong 266000,China)
出处
《通信技术》
2021年第6期1384-1391,共8页
Communications Technology