合成孔径雷达(Synthetic aperture radar,SAR)图像因为相干斑现象和目标响应的空间变化呈现出一种纹理特性,局部二进编码等局部图像特征在光学纹理描述中获得较好的结果,但光学纹理特征在描述SAR图像纹理特性中因为相干成像特性往往失效...合成孔径雷达(Synthetic aperture radar,SAR)图像因为相干斑现象和目标响应的空间变化呈现出一种纹理特性,局部二进编码等局部图像特征在光学纹理描述中获得较好的结果,但光学纹理特征在描述SAR图像纹理特性中因为相干成像特性往往失效.本文在前期工作纹理特征框架的基础上,提出了一种局部重要性采样二进编码的SAR图像纹理特征(Feature extraction based on local important sampling binary,LISBF)描述方法:首先,利用样本图像对局部采样位置进行随机自适应采样,基于重要性采样(Important sample,IS)方法输出递归学习位置结果;然后,利用学习出的纹理重要采样点对进行二进特征编码;最后,通过映射和统计生成描述算子.该特征较固定位置采样能够获取更大范围信息,同时能通过采样避免特征维数的急剧增大;通过自适应学习重要性关键点较随机采样更容易捕捉纹理固有信息;较好地适应了SAR图像极低信噪比和斑点现象的纹理.本文将该特征用于真实图像和标准纹理库的分类研究,实验结果证明了该特征的有效性.展开更多
Texture synthesis has been developed for several years.The traditional technique can generate a larger image from a small image while avoid feeling of repetition or uncontinuity.Some constrained synthesis methods whic...Texture synthesis has been developed for several years.The traditional technique can generate a larger image from a small image while avoid feeling of repetition or uncontinuity.Some constrained synthesis methods which can synthesize image according to special location demand or other demands have been also proposed in recent years.However,in general,these constrained texture synthesis methods are simple and have few controllable factors to meet user＇s diverse needs.To control multiple-sample texture synthesis more flexibly in various aspects such as synthesis location,proportion and semantic objects,we present an interactive texture synthesis approach based on circular patches and constrained by objects according to a certain ratio.With this approach,source exemplars and the target image are firstly divided into several regions with different characters.Users can click the blocks in the source exemplars and the want-to-be-synthesized region in the target image,and then texture in the target image is synthesized with the corresponding regions in the source exemplars.In the process of texture synthesis,circular patch instead of square patch is used to eliminate the aliasing phenomena.Images are synthesized from multiple sample images with ratio constraint and experiments on images show that our approach can get effective results of ratio-constrained multi-sample synthesis.展开更多
目的针对基于SURF特征点的图像配准算法对颜色单一的彩色图像提取的特征点较少及配准时间复杂度高等问题,提出一种基于融合特征的快速SURF(speed up robust features)配准算法。方法该算法首先提取图像的颜色不变量边缘特征和CS-LBP(cen...目的针对基于SURF特征点的图像配准算法对颜色单一的彩色图像提取的特征点较少及配准时间复杂度高等问题,提出一种基于融合特征的快速SURF(speed up robust features)配准算法。方法该算法首先提取图像的颜色不变量边缘特征和CS-LBP(central symmetry-local binary patterns)纹理特征形成融合特征灰度图,并利用颜色直方图的方差自适应调节融合特征间的权重。其次,在融合特征灰度图上提取SURF(speed up robust features)特征点及描述子。再次,用最近邻匹配法形成粗匹配对,结合改进的快速RANSAC(random sample consensus)算法得到精匹配对。最后,使用最小二乘法求出映射关系用于配准图像。结果本文算法能够在融合特征上提取更稳定的SURF特征点,用该特征点进行配准能提高配准5%精度,且减少时间复杂度15%,实现了对普通场景下图像的快速配准。结论本文算法能提取稳定数量的特征点,提高了精确度与鲁棒性,并通过改进的RANSAC算法提高了执行效率,降低了迭代次数。展开更多
文摘合成孔径雷达(Synthetic aperture radar,SAR)图像因为相干斑现象和目标响应的空间变化呈现出一种纹理特性,局部二进编码等局部图像特征在光学纹理描述中获得较好的结果,但光学纹理特征在描述SAR图像纹理特性中因为相干成像特性往往失效.本文在前期工作纹理特征框架的基础上,提出了一种局部重要性采样二进编码的SAR图像纹理特征(Feature extraction based on local important sampling binary,LISBF)描述方法:首先,利用样本图像对局部采样位置进行随机自适应采样,基于重要性采样(Important sample,IS)方法输出递归学习位置结果;然后,利用学习出的纹理重要采样点对进行二进特征编码;最后,通过映射和统计生成描述算子.该特征较固定位置采样能够获取更大范围信息,同时能通过采样避免特征维数的急剧增大;通过自适应学习重要性关键点较随机采样更容易捕捉纹理固有信息;较好地适应了SAR图像极低信噪比和斑点现象的纹理.本文将该特征用于真实图像和标准纹理库的分类研究,实验结果证明了该特征的有效性.
基金Supported by the National Natural Science Foundation of China(60533080)
文摘Texture synthesis has been developed for several years.The traditional technique can generate a larger image from a small image while avoid feeling of repetition or uncontinuity.Some constrained synthesis methods which can synthesize image according to special location demand or other demands have been also proposed in recent years.However,in general,these constrained texture synthesis methods are simple and have few controllable factors to meet user＇s diverse needs.To control multiple-sample texture synthesis more flexibly in various aspects such as synthesis location,proportion and semantic objects,we present an interactive texture synthesis approach based on circular patches and constrained by objects according to a certain ratio.With this approach,source exemplars and the target image are firstly divided into several regions with different characters.Users can click the blocks in the source exemplars and the want-to-be-synthesized region in the target image,and then texture in the target image is synthesized with the corresponding regions in the source exemplars.In the process of texture synthesis,circular patch instead of square patch is used to eliminate the aliasing phenomena.Images are synthesized from multiple sample images with ratio constraint and experiments on images show that our approach can get effective results of ratio-constrained multi-sample synthesis.
