光谱学与光谱分析, 2020, 40 (2): 586, 网络出版: 2020-05-12   

红外偏振图像的舰船目标检测

Ship Target Detection Based on Infrared Polarization Image
作者单位
海军航空大学, 山东 烟台 264000
摘要
红外成像系统作为探测舰船目标的一种重要方式, 在**侦察中起着至关重要的作用, 面对复杂背景、 恶劣天气环境等情况时, 目标与背景局部对比度较低导致红外系统的探测准确率、 查全率等性能指标受到严重影响, 针对上述问题, 开展了基于红外偏振图像的舰船目标检测方法研究。 通过8~12 μm长波波段红外偏振图像采集系统实际采集86组4个偏振方向(0°, 45°, 90°, 135°)的红外偏振图像, 样本中舰船目标309个。 对同场景下不同偏振方向的红外偏振图像及红外强度/偏振度图像的目标与背景局部对比度计算, 发现海面与舰船目标偏振特征差异能够有效提高目标与背景局部对比度。 在前视红外图像中, 舰船目标通常位于海天线附近或下方, 但复杂背景及天气等因素干扰对红外图像中检测海天线影响较大, 为此提出红外偏振图像海天线检测方法, 对红外偏振图像直方图进行高斯滤波消除局部极值, 依据海面与背景的偏振特征差异, 利用双峰法阈值分割检测海天线, 最后利用霍夫变换检测海天线, 分割出海面作为目标候选区域。 针对红外偏振图像受到海杂波严重干扰的问题, 提出海杂波背景抑制算法, 采取背景抑制, 距离加权方法抑制偏振图像中杂乱的海杂波背景。 最后, 利用MSER算法检测舰船目标, 根据舰船目标特性约束条件剔除非目标区域。 对86组红外偏振图像进行舰船目标检测实验, 所提出的方法能够有效克服复杂背景和海杂波等因素的干扰准确检测海天线, 通过海杂波背景抑制及舰船特征约束的方法消除海杂波及海岸对舰船检测所带来的干扰, 检测准确率、 查全率分别为93.2%和95.7%, 优于红外舰船目标检测方法, 特别在红外图像对比度低的场景下检测效果提升明显, 检测准确率、 查全率分别提高了46.5%和16.4%。 表明充分考虑红外偏振图像中舰船目标与背景的偏振特征差异能够有效提高目标与背景的局部对比度, 有利于准确检测海天线, 提高舰船检测准确率和查全率, 对复杂背景及恶劣天气有较强的适应性, 在**应用中具有极大应用价值, 对红外波段舰船目标检测技术的发展有着极为重要的意义。
Abstract
As an important method detecting ship targets, infrared imaging system plays a vital role in military reconnaissance. In the face of complex background, bad weather environment, etc., the local contrast of target and background is low, which causes that the performance indicators of infrared system such as detection accuracy and recall rate are seriously affected. Considering the above problems, the research on ship target detection method based on infrared polarization image is carried out. Through the long-wave uncooled infrared polarization image acquisition system, 86 sets of infrared polarization images with 4 polarization directions (0°, 45°, 90°, 135° ) were collected, and 309 ship targets were sampled. For the infrared polarization image with different polarization directions, the target and background local contrast of the infrared intensity/polarization image in the same scene, it was found that the difference between the sea surface and the ship target polarization characteristics can effectively improve the target and background local contrast. In the forward-looking infrared image, the ship target is usually located near or below the sea-sky line, but the complex background and weather and other factors have a great influence on the detection of the sea-sky line in the infrared image. For this reason, the infrared polarization image sea-sky line detection method is proposed. Gaussian filtering is used to eliminate the local extremum in the histogram of polarization image. According to the difference of the polarization characteristics between the sea and the background, the sea-sky line is detected by the bimodal method threshold segmentation. Finally, the sea-sky line is detected by the Hough transform, and the sea surface is segmented as the target candidate region. Aiming at the problem that the infrared polarization image is seriously interfered by sea clutter, the sea clutter background suppression algorithm is proposed. The background suppression and distance weighting method are used to suppress the complicated sea clutter background in the polarization image. Finally, the ship targets are detected according to the MSER algorithm, after they are constrained basing on the characteristics of the ship target. The ship target detection experiment is carried out on 86 sets of infrared polarization images. The proposed method can effectively overcome the interference of complex background, sea clutter and other factors to accurately detect the sea-sky line, and the detection precision ratio and the recall ratio are 93.2% and 95.7%, respectively, which is better than the infrared ship target detection method, especially in the scene with the low contrast of infrared image. Obviously, the detection precision ratio and the recall ratio increase by 46.5% and 16.4%, respectively. The research results show that considering the difference of polarization characteristics between the ship target and the background in the infrared polarization image can effectively improve the local contrast between the target and the background, which is conducive to accurately detecting the sea-sky line, improving the ship detection precision ratio and the recall ratio, and complex background. It has strong adaptability to complex background and bad weather, which has great application value in military applications. This study has great significance for the development of ship target detection technology in the infrared band.

宫剑, 吕俊伟, 刘亮, 仇荣超. 红外偏振图像的舰船目标检测[J]. 光谱学与光谱分析, 2020, 40(2): 586. GONG Jian, L Jun-wei, LIU Liang, QIU Rong-chao. Ship Target Detection Based on Infrared Polarization Image[J]. Spectroscopy and Spectral Analysis, 2020, 40(2): 586.

本文已被 6 篇论文引用
被引统计数据来源于中国光学期刊网
引用该论文: TXT   |   EndNote

相关论文

加载中...

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!