高分辨率遥感影像的目标分类与识别,是对地观测系统进行图像分析理解,以及自动目标识别系统提取目标信息的重要手段。本文综述了当前国内外在可见光、红外、合成孔径雷达和合成孔径声纳等遥感影像的目标分类与识别的关键技术和最新研究进展。首先,讨论了高分辨率遥感影像的目标分类与识别问题的主要研究层次和内容;其次,深入分析了高分辨率遥感影像目标分类与识别,在滤波降噪、特征提取、目标检测、场景分类、目标分类和目标识别的关键技术及其所存在的问题;最后,结合并行计算、神经计算和认知计算等技术,讨论了目标分类与识别的可行性方案。具体包括：（1）高性能并行计算在高分辨率遥感图像处理的主流技术,并给出了基于Hadoop+OpenMP+CUDA的高分辨率遥感影像混合并行处理架构;（2）深度学习对于提升目标分类和识别精度的应用前景,以及基于深度神经网络的多层次遥感影像目标识别方法;（3）认知计算在解决遥感影像大数据不确定性分析的模型与算法,并讨论了层次主题模型的多尺度遥感影像场景描述方案。此外,根据媒体神经认知计算的相关研究,探讨了遥感影像大数据的目标分类和识别的发展趋势和研究方向。

<p>给出一种从高分辨率遥感影像中提取道路的方法。先对遥感影像进行滤波预处理获取纹理特征；再以图割解算经典能量函数获取道路图斑，其中利用高斯混合模型构建T链；接着利用直线匹配法获得初始道路段；最后用图割解算融合了两种形状先验的能量函数，结合动态外推的思路连接道路段。对不同遥感影像的实验证明了算法的有效性。</p>

<p>针对遥感影像道路提取问题,探索了一种基于道路条带自动检测道路中心线的方法。首先基于概率增进树算法获取道路候选点,并通过形态学运算得到光滑和完整的道路条带; 然后结合细化运算自动检测道路中心线。但检测的中心线存在&quot;毛刺&quot;且局部曲率变化过大,不符合道路的形状特征。针对该问题,引入了测地距离理论,并用迭代方法去除&quot;毛刺&quot;,获得初始道路中心线; 再通过Dijkstra最短路径算法优化初始结果; 最后根据方向一致性和道路连续性获取最终的道路中心线。采用高分辨率航空影像对上述方法的实验结果证明了该方法的有效性和可行性。</p>

基于灰度形态学,提出一种从高分辨率遥感图像提取道路网络的方法.首先利用灰度形态特征对遥感影像进行分割,进而得到基本的道路网络轮廓.然后在此基础上,利用线段特征匹配方法提取道路网络.提出的方法能适应于从道路和背景区别不很清楚的遥感图像中提取道路.实验结果也表明,本文方法能有效地从遥感影像中提取道路网络.

In this work we present a novel vision-based system for automatic detection and extraction of complex road networks from various sensor resources such as aerial photographs, satellite images, and LiDAR. Uniquely, the proposed system is an integrated solution that merges the power of perceptual grouping theory (Gabor filtering, tensor voting) and optimized segmentation techniques (global optimization using graph-cuts) into a unified framework to address the challenging problems of geospatial feature detection and classification. Firstly, the local precision of the Gabor filters is combined with the global context of the tensor voting to produce accurate classification of the geospatial features. In addition, the tensorial representation used for the encoding of the data eliminates the need for any thresholds, therefore removing any data dependencies. Secondly, a novel orientation-based segmentation is presented which incorporates the classification of the perceptual grouping, and results in segmentations with better defined boundaries and continuous linear segments. Finally, a set of gaussian-based filters are applied to automatically extract centerline information (magnitude, width and orientation). This information is then used for creating road segments and transforming them to their polygonal representations.

This paper surveys the state of the art on automatic road extraction for GIS update from aerial and satellite imagery. It presents a bibliography of nearly 250 references related to this topic. The work includes main approaches on general methods of road network extraction and reconstruction, road tracking methods, morphological analysis, dynamic programming and snakes, methods multi-scale and multi-resolution, stereoscopic and multi-temporal analysis, hyper-spectral experiments, and other techniques for road extraction. Likewise, other approaches related in any way with the road extraction topic are also considered. Between them different papers on segmentation, vectorization, optimization, evaluation, semantic nets and neural networks, fusion techniques, fuzzy logic, and other methods are discussed. A novel classification of road extraction methods according to our criteria is included in order to provide a significant contribution to research in this topic.

In this paper, a unique approach for road extraction utilizing pixel spectral information for classification and image segmentation-derived object features was developed. In this approach, road extraction was performed in two steps. In the first step, support vector machine (SVM) was employed merely to classify the image into two groups of categories: a road group and a non-road group. For this classification, support vector machine (SVM) achieved higher accuracy than Gaussian maximum likelihood (GML). In the second step, the road group image was segmented into geometrically homogeneous objects using a region growing technique based on a similarity criterion, with higher weighting on shape factors over spectral criteria. A simple thresholding on the shape index and density features derived from these objects was performed to extract road features, which were further processed by thinning and vectorization to obtain rood centerlines. The experiment showed the proposed approach worked well with images comprised by both rural and urban area features.

This letter presents a two-step method for urban main road extraction from high-resolution remotely sensed imagery by integrating spectral-spatial classification and shape features. In the first step, spectral-spatial classification segments the imagery into two classes, i.e., the road class and the nonroad class, using path openings and closings. The local homogeneity of the gray values obtained by local Geary's C is then fused with the road class. In the second step, the road class is refined by using shape features. The experimental results indicated that the proposed method was able to achieve a comparatively good performance in urban main road extraction.

This study presents an automatic method for accurate road centreline extraction from classified images. Unlike earlier methods that have mainly relied on the thinning algorithm, the proposed automatic method selects end points from a classified image and then connects them using the geodesic method to formulate the road network. As the proposed method generates the road centreline by tracing along the ridge line of the optimally oriented flux (OOF) measure, this method is able to eliminate undesired spurs while retaining the correct road network topology. Three test images are used to quantitatively evaluate the proposed method. Experiments show that the proposed method yields a substantial improvement over the thinning algorithm.

基于模板匹配的道路追踪方法是道路提取中较实用的一类方法,但传统模板匹配方法主要以相关系数作为相似性测度,对车辆、树荫等遮挡敏感,不适用于高分辨率遥感影像道路提取。针对这一问题,本文采用一种稳健的相似性测度,设计了一种基于均值漂移的道路中心点匹配算法,克服了传统模板匹配对遮挡敏感的缺点;然后运用卡尔曼滤波,实现高分辨率遥感影像道路中心线追踪。试验表明,该方法能够准确提取高分辨率遥感影像道路中心线,对车辆、树荫等遮挡具有稳健性。

Abstract We present a unified computational framework which properly implements the smoothness constraint to gen- erate descriptions in terms of surfaces, regions, curves, and labelled junctions, from sparse, noisy, binary data in 2-D or 3-D. Each input site can be a point, a point with an associated tangent direction, a point with an associated normal direction, or any combination of the above. The methodology is grounded on two elements: tensor calculus for representation, and linear voting for communication: each input site communicates its infor- mation (a tensor) to its neighborhood through a pre- defined (tensor) field, and therefore casts a (tensor) vote. Each site collects all the votes cast at its location and encodes them into a new tensor. A local, parallel marching process then simultaneously detects features. The proposed approach is very different from traditional variational approaches, as it is non-iterative. Further- more, the only free parameter is the size of the neigh- borhood, related to the scale. We have developed several algorithms based on the proposed methodology to address a number of early vision problems, including perceptual grouping in 2-D and 3-D, shape from stereo, and motion grouping and segmentation, and the results are very encouraging.