2.118

影响因子

    高级检索

    一种基于差分滤波的红外舰船目标检测方法

    Infrared Ship Targets Detection Method Based on Di erential Filtering

    • 摘要: 无人机有时需要在缺乏预先红外样本图像的情况下对指定区域的舰船目标进行侦察监视. 针对该情况下的红外舰船目标自动检测问题, 设计了一种基于多分辨率差分滤波的检测方法. 该方法对输入图像在多个分辨率下作差分滤波, 再将滤波结果与选择性搜索算法结合, 得到检测结果. 实验中, 选择139 张图像作为测试图像数据集. 使用多分辨率差分滤波检测方法对测试图像数据集进行红外舰船目标检测, 平均检测准确率达到89.378 %, 比采用HOG + SVM 检测方法高8.445 %.

       

      Abstract: For that UAV reconnoiters and monitors ship targets in the designated area when there is no e ective pre sample image accumulation, an infrared ship target detection method based on multi-resolution di erential filtering is designed. In this method, the input images are filtered in di erent resolutions, and then detection results are obtained by combining the filtering results with the selective search algorithm. In the experiment, 139 images are selected as the test image data set, and the multi-resolution di erence filtering method is used to detect infrared ship targets in it. The average detection accuracy is 89.378 %, which is 8.445% higher than that of the HOG + SVM method.

       

    /

    返回文章
    返回