基于视觉注意的SAR目标快速检测算法研究

发布时间：2018-05-04 04:15

本文选题：SAR图像 + 目标检测　；参考：《电子科技大学》2015年硕士论文

【摘要】：合成孔径雷达(Synthetic Aperture Radar,SAR)的工作不受时间和天气的影响,并且具有很强的透射性,凭借这些优势,它在军事领域占据重要地位,军事作战提供军事作战所需要的侦察和目标探测信息,同时也普遍应用于气象预告、地形地貌研究、道路检测、灾情监控等民用方面。SAR目标检测是利用目标的灰度、纹理、形状、边缘和方向等信息在SAR图像中确定其位置,将目标与背景分离的图像处理技术。随着SAR图像分辨率不断提高,目标信息呈现爆炸性增长,如何快速有效地对SAR目标进行检测,是目前SAR图像处理领域研究的热点问题。视觉注意机制是指人眼对外界环境中的海量信息进行筛选处理,并只对有用信息做出响应的信息处理机制。视觉注意计算模型是利用数学建模的方法,结合计算机提供的仿真环境,模拟人眼视觉感知系统对信息的处理过程,在图像分析处理领域,视觉注意机制掀起了研究热潮。本文将视觉注意机制引入到SAR目标检测领域,论文的主要研究内容和创新之处如下:(1)研究了海内外较为成熟的视觉注意计算模型,并与人类视觉感知系统的工作过程相结合,对应用较广的Itti经典模型和Hou谱残差模型进行了详细的分析。阐述了两种处理方式在SAR图像处理方面的不足。(2)研究了一种适用于SAR图像的视觉注意模型。该模型建立在Itti经典模型框架的基础之上,提取能够表征SAR目标的特征,通过特征整合机制得到SAR图像的显著图。仿真对比实验的结果表明,由该模型得到的显著图中目标轮廓形状清晰,定位准确,且计算速度较快。(3)研究了一种基于视觉显著性的SAR目标快速检测新算法。该算法首先利用视觉注意模型得到输入图像的显著图,然后结合SAR目标的尺寸信息对显著区域进行高斯平滑和筛选,然后经过感兴趣区域(Region of Interest,ROI)矩形补齐和配准处理,完成对整幅图像的目标检测。该方法解决问题的思路较为新颖,结合了视觉显著性,仿真实验结果表明与传统的检测算法相比,该算法具有快速准确的优点。
[Abstract]:The work of synthetic Aperture radar (SAR) is not affected by time and weather, and it has strong transmissibility. With these advantages, it plays an important role in the military field. Military operations provide reconnaissance and target detection information needed for military operations. At the same time, they are also widely used in civilian aspects such as weather forecasting, terrain and geomorphology research, road detection, disaster monitoring and other civilian aspects. SAR target detection is based on the grayscale and texture of the target. Shape, edge and direction information in the SAR image to determine the location of the image processing technology to separate the target from the background. With the improvement of the resolution of SAR images and the explosive growth of target information, how to detect SAR targets quickly and effectively is a hot issue in the field of SAR image processing. Visual attention mechanism is an information processing mechanism in which the human eye filters and processes massive information in the external environment and only responds to useful information. Visual attention computing model is to simulate the processing process of human visual perception system by using mathematical modeling method and computer simulation environment. In the field of image analysis and processing, visual attention mechanism has aroused a hot research. In this paper, the visual attention mechanism is introduced into the field of SAR target detection. The main contents and innovations of this paper are as follows: 1) the more mature visual attention computing model at home and abroad is studied and combined with the working process of human visual perception system. The classical Itti model and the Hou spectral residual model are analyzed in detail. The deficiency of two processing methods in SAR image processing is discussed. A visual attention model suitable for SAR image is studied. The model is based on the framework of the classical Itti model and extracts the features that can represent the SAR target. The salient map of the SAR image is obtained by the feature integration mechanism. The simulation results show that the contour of the target is clear, the location is accurate, and the computing speed is faster. (3) A new algorithm for fast detection of SAR targets based on visual saliency is studied. Firstly, the saliency map of the input image is obtained by using visual attention model, then the salient region is smoothed and filtered by Gao Si with the size information of the SAR target, and then the region of interest is patched and registered by rectangle. Complete the target detection of the whole image. The method is novel in solving the problem and combines visual salience. The simulation results show that the algorithm has the advantages of fast and accurate compared with the traditional detection algorithm.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN957.52

【参考文献】