采用塔式关键词直方图的舰船图像实时分类检测方法

发布时间：2018-03-16 03:34

本文选题：舰船图像分类　切入点：塔式关键词直方图　出处：《科学技术与工程》2017年33期 　论文类型：期刊论文

【摘要】：针对传统舰船分类检测方法实时性差、容易受到物理噪声干扰等问题,采用塔式关键词直方图和支持向量机的检测方法对不同类别水面舰船图像进行实时分类检测。通过对不同类别的舰船图像进行分类实验,进一步综合确定适合的塔式关键词描述子参数及支持向量机核函数参数。实验结果表明,舰船分类检测准确率较已有检测方法有所提高。基于塔式关键词直方图和支持向量机的检测方法能够实现可靠、实时的舰船图像分类检测。
[Abstract]:Aiming at the problem that the traditional ship classification detection method is poor in real time and easily disturbed by physical noise, The methods of tower keyword histogram and support vector machine are used to detect different kinds of ship images in real time. Furthermore, the suitable parameters of tower keyword description subparameter and support vector machine kernel function are determined synthetically. The experimental results show that, The accuracy of ship classification detection is higher than that of existing methods, and the detection method based on tower keyword histogram and support vector machine can achieve reliable and real-time ship image classification detection.
【作者单位】：海军大连舰艇学院;
【基金】：十三五国防预研项目(30203010303)资助
【分类号】：TP391.41

【相似文献】