基于小面核滤波的快速红外舰船检测方法

发布时间：2018-10-08 21:53

【摘要】：针对复杂海天背景下红外舰船目标的实时检测需求,提出了一种基于小面核滤波的快速红外舰船检测方法.在频域视觉显著性计算原理的启发下,结合小面拟合模型,设计了一种高效的时域小面卷积核,以抑制背景并增强具有高亮灰度对比边缘的红外舰船目标,进而确定候选目标区域.然后采用一种L0梯度最小化滤波方法,实现候选区域中目标区域的灰度均衡,以完整并准确提取目标形状.最后根据舰船的尺度和形状特性,完成目标判别.实验中对3组公开实拍红外图像数据集进行了测试,实验结果表明:本方法具有较好的鲁棒性,其对3组数据的检测准确率均高于90%,在PC平台上的处理速度维持在10ms量级,优于现有的典型方法,能够满足多数工程应用需求.
[Abstract]:A fast infrared ship detection method based on small surface kernel filter is proposed to meet the requirement of real-time detection of infrared ship targets under complex sea and sky background. Inspired by the principle of visual salience calculation in frequency domain, an efficient time domain small surface convolution kernel is designed in combination with the small plane fitting model to suppress the background and enhance the infrared ship target with highlight gray contrast edge. Then the candidate target area is determined. Then, an L0 gradient minimization filtering method is used to realize the gray level equalization of the target region in the candidate region, and the shape of the target is extracted completely and accurately. Finally, according to the scale and shape characteristics of the ship, the target identification is completed. Three groups of real infrared image data sets are tested in the experiment. The experimental results show that this method has good robustness, the accuracy of the three groups of data detection is higher than 90 pieces, and the processing speed on PC platform is in the order of 10ms. Better than the existing typical methods, can meet the needs of most engineering applications.
【作者单位】：中国航空无线电电子研究所航空电子系统综合技术重点实验室;
【基金】：国家自然科学基金资助项目(2014CB744900)
【分类号】：E925;TP391.41

【相似文献】