一种针对海面SAR图像的视觉注意模型设计
|
摘要
在研究了经典ITTI等视觉注意模型的理论基础上,结合海面SAR图像背景及目标特点,对传统视觉模型应用于海面SAR图像的缺陷进行分析总结,提出一种适用于海面SAR图像视觉注意模型设计算法。首先,模型借鉴经典ITTI模型的基本框架,选择并提取了能够较好描述SAR图像的纹理和形状特征,求取相应的特征显著图;其次,采用新的特征显著图整合机制替代经典模型的线性相加机制进行显著图融合得到总显著图;最后,综合各特征显著图下注意焦点的灰度特征,选择最佳的显著性表征,完成通过多尺度竞争策略对显著图的滤波及阈值分割实现显著区域的精确筛选,从而完成SAR图像的显著区域检测。实验采用Terra SAR-X等多幅卫星数据进行仿真实验,结果验证了模型良好的显著性检测效果,更符合实际高分辨率图像目标检测的应用需求。通过进一步与经典视觉模型对比分析,模型在改善了由斑点噪声和不均匀的海杂波背景对检测结果产生的虚警影响的同时,检测速度也较之提高了25%~45%。
On the basis of studying the theories of classical ITTI visual attention models,the defects of traditional visual models applied to sea-surface SAR images are summarized according to the characteristics of the background and the target of sea-surface SAR images. A visual attention model design algorithm for seasurface SAR images is proposed. Firstly,the model uses the basic framework of the classical ITTI model,selects and extracts the texture and shape features that can describe the SAR image well. Then the corresponding saliency map of features is obtained. Secondly,the new integration mechanism of the saliency map of features is adopted to replace the linear-adding mechanism of the classical model for fusing the saliency maps and obtaining the overall saliency map. Finally,the gray features of the attention focus of all the saliency maps are integrated to select the optimal significance characterization. By using the multi-scale competitive strategy,the filtering and threshold segmentation are completed to realize the accurate screening of significant areas. Therefore,the detection of the significant areas of SAR images is completed. Experiments were carried out by using Terra-SAR-X and other satellite data,and their results verified the good significancedetection effects of the model. The model can better meet the demands of the detection of high-resolution image targets. By carrying out further comparative analysis with the classical visual model,it is discovered that the proposed algorithm can not only reduce the impact of the false alarm caused by speckle noise and uneven sea-clutter background on the detection result,but also greatly improve the detection speed by 25% to 45%.
On the basis of studying the theories of classical ITTI visual attention models,the defects of traditional visual models applied to sea-surface SAR images are summarized according to the characteristics of the background and the target of sea-surface SAR images. A visual attention model design algorithm for seasurface SAR images is proposed. Firstly,the model uses the basic framework of the classical ITTI model,selects and extracts the texture and shape features that can describe the SAR image well. Then the corresponding saliency map of features is obtained. Secondly,the new integration mechanism of the saliency map of features is adopted to replace the linear-adding mechanism of the classical model for fusing the saliency maps and obtaining the overall saliency map. Finally,the gray features of the attention focus of all the saliency maps are integrated to select the optimal significance characterization. By using the multi-scale competitive strategy,the filtering and threshold segmentation are completed to realize the accurate screening of significant areas. Therefore,the detection of the significant areas of SAR images is completed. Experiments were carried out by using Terra-SAR-X and other satellite data,and their results verified the good significancedetection effects of the model. The model can better meet the demands of the detection of high-resolution image targets. By carrying out further comparative analysis with the classical visual model,it is discovered that the proposed algorithm can not only reduce the impact of the false alarm caused by speckle noise and uneven sea-clutter background on the detection result,but also greatly improve the detection speed by 25% to 45%.
引文
[1]MOREIRA A,PRATS-IRAOLA P,YOUNIS M,et al.A tutorial on synthetic aperture radar[J].IEEE Geoscience and Remote Sensing Magazine,2013,1(1):6-43.
[2]邓云凯,赵凤军,王宇.星载SAR技术的发展趋势及应用浅析[J].雷达学报,2012,1(1):1-10.
[3]邢相薇,计科峰,康利鸿,等.HRWS SAR图像舰船目标监视技术研究综述[J].雷达学报,2015,4(1):107-121.
[4]ITTI L,KOCH C,NIEBUR E.A model of saliency-based visual attention for rapid scene analysis[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1998,20(11):1254-1259.
[5]HOU X D,ZHANG L Q.Saliency detection:a spectral residual approach[C]//IEEE Conference on Computer Vision and Pattern Recognition,2007:1-8.
[6]GUO C L,MA Q,ZHANG L M.Spatio-temporal saliency detection using phase spectrum of quaternion Fourier transform[C]//IEEE Conference on Computer Vision and Pattern Recognition,2008:1-8.
[7]YU Y,WANG B,ZHANG L M.Pulse discrete cosine transform for saliency-based visual attention[C]//The 8th Conference on Development and Learning,2009:1-6.
[8]ITTI L,KOCH C,BRAUN J.Revisiting spatial vision:toward a unifying model[J].Journal of the Optical Society of America,2000,17(11):1899-1917.
[9]OLIVA A,TORRALBA A.Modeling the shape of the Scene:a holistic representation of the spatial envelope[J].International Journal of Computer Vision,2001,42(3):145-175.
[10]GAO D S,HAN S,VASCONCELOS N.Discriminant saliency,the detection of suspicious coincidences,and applications to visual recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2009,31(6):989-1005.
[11]张志龙,杨卫平,张焱,等.基于频谱残留变换的红外遥感图像舰船目标检测方法[J].电子与信息学报,2015,37(9):2144-2150.
[12]ITTI L.Models of bottom-up and top-down visual attention[D].California:California Institute of Technology,2000.
[2]邓云凯,赵凤军,王宇.星载SAR技术的发展趋势及应用浅析[J].雷达学报,2012,1(1):1-10.
[3]邢相薇,计科峰,康利鸿,等.HRWS SAR图像舰船目标监视技术研究综述[J].雷达学报,2015,4(1):107-121.
[4]ITTI L,KOCH C,NIEBUR E.A model of saliency-based visual attention for rapid scene analysis[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1998,20(11):1254-1259.
[5]HOU X D,ZHANG L Q.Saliency detection:a spectral residual approach[C]//IEEE Conference on Computer Vision and Pattern Recognition,2007:1-8.
[6]GUO C L,MA Q,ZHANG L M.Spatio-temporal saliency detection using phase spectrum of quaternion Fourier transform[C]//IEEE Conference on Computer Vision and Pattern Recognition,2008:1-8.
[7]YU Y,WANG B,ZHANG L M.Pulse discrete cosine transform for saliency-based visual attention[C]//The 8th Conference on Development and Learning,2009:1-6.
[8]ITTI L,KOCH C,BRAUN J.Revisiting spatial vision:toward a unifying model[J].Journal of the Optical Society of America,2000,17(11):1899-1917.
[9]OLIVA A,TORRALBA A.Modeling the shape of the Scene:a holistic representation of the spatial envelope[J].International Journal of Computer Vision,2001,42(3):145-175.
[10]GAO D S,HAN S,VASCONCELOS N.Discriminant saliency,the detection of suspicious coincidences,and applications to visual recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2009,31(6):989-1005.
[11]张志龙,杨卫平,张焱,等.基于频谱残留变换的红外遥感图像舰船目标检测方法[J].电子与信息学报,2015,37(9):2144-2150.
[12]ITTI L.Models of bottom-up and top-down visual attention[D].California:California Institute of Technology,2000.