摘要：【中文】本发明公开了一种面向三维场景的矢量图标绘制方法及装置，本发明的矢量图标绘制方法包括：1)各种图标对象(规则/不规则)的定义；2)设计实现各图标对象的几何算法；3)绘制图标对像，保存其基本信息；4)针对各种图标对象进行编辑逻辑的设定。本发明包含规则和非规则两类图标绘制及编辑，可灵活扩展，适用于各类专业组合图元的绘制和编辑。本发明适用于桌面端、Web端、移动端各类型图标绘制场景，已在各端应用中实践。同时该方案虽为三维绘制，但可灵活扩展，适用于各类专业组合图元的绘制和编辑，亦可稍作调整适用于二维图标的绘制和编辑。该图标绘制算法成熟、稳定，已通过各大开源地图平台实际应用予以验证，具备较大推广价值。 【EN】The invention discloses a vector icon drawing method and a device for three-dimensional scenes, wherein the vector icon drawing method comprises the following steps: 1) definition of various icon objects (regular/irregular); 2) designing a geometric algorithm for realizing each icon object; 3) drawing the icon object and storing the basic information of the icon object; 4) the setting of the edit logic is performed for each icon object. The method comprises regular icon drawing and editing and irregular icon drawing and editing, can be flexibly expanded, and is suitable for drawing and editing various professional combined primitives. The method is suitable for drawing scenes of various types of icons at a desktop end, a Web end and a mobile end, and is practiced in application of each end. Meanwhile, the scheme is three-dimensional drawing, but can be flexibly expanded, is suitable for drawing and editing various professional combined primitives, and can also be slightly adjusted and suitable for drawing and editing two-dimensional icons. The icon drawing algorithm is mature and stable, is verified by practical application of each large-open-source map platform, and has a great popularization value.

摘要：【中文】本发明提出了一种通过抑制非感兴趣信息的语义分割方法、设备及存储设备，本发明基于深度学习库优化神经网络，提高语义分割结果的精度，主要包括以下步骤：1)构建基础Unet模型；2)添加注意力机制；3)门特征图与当前层结果相乘；4)添加新输出结果和多损失函数；5)对待进行语义分割的图像进行图像语义分割。本方法可以提高语义分割神经网络的精度并有效抑制非感兴趣信息。 【EN】The invention provides a semantic segmentation method, equipment and storage equipment through inhibiting non-interesting information, wherein a neural network is optimized based on a deep learning library, the precision of a semantic segmentation result is improved, and the semantic segmentation method mainly comprises the following steps: 1) constructing a basic Unet model; 2) adding an attention mechanism; 3) multiplying the gate feature graph by the current layer result; 4) adding a new output result and a multi-loss function; 5) and performing image semantic segmentation on the image to be subjected to the semantic segmentation. The method can improve the precision of the semantic segmentation neural network and effectively inhibit the non-interesting information.

摘要：【中文】本发明公开了一种可扩展的适应二三维场景的矢量标绘图形态势推演方法。矢量标绘图形态势推演与仿真可提供丰富直观的战场态势表达和有序的态势推演过程，有助于指挥人员快速制定作战计划、赢得作战时间，从而对获取作战的胜利具有重大的指导意义。本发明主要包含如下几个关键步骤：设计各类矢量标绘图形态势规则；设计实现矢量标绘图形及态势文件管理；编辑矢量标绘图形态势；实现矢量标绘图形态势模拟。本发明适用于桌面端、Web端、移动端各端矢量标绘图形态势推演应用，可应用到二维地图场景和三维地图场景中，而且本发明设计上具有良好的扩展性，支持其它矢量标绘图形类型的扩展；本发明可灵活扩展，适用于各类专业组合图元动画类方案制作。 【EN】The invention discloses an extensible vector plotting graph situation deduction method adaptive to two-dimensional and three-dimensional scenes. The vector plotting graph situation deduction and simulation can provide rich and visual battlefield situation expression and an ordered situation deduction process, and is beneficial to commanders to quickly make a battle plan and win the battle time, so that the vector plotting graph situation deduction and simulation has great guiding significance for obtaining the victory of the battle. The invention mainly comprises the following key steps: designing various vector plotting graph situation rules; designing and realizing vector plotting graphs and situation file management; editing vector plotting graph situations; and realizing vector plotting graphic situation simulation. The method is suitable for the vector plotting graph situation deduction application of each end of a desktop end, a Web end and a mobile end, can be applied to a two-dimensional map scene and a three-dimensional map scene, has good expansibility in design, and supports the expansion of other vector plotting graph types; the method can be flexibly expanded and is suitable for the production of various professional combined primitive animation schemes.

摘要：【中文】本发明公开了一种基于深度卷积神经网络的遥感影像红树林提取方法及系统，本发明首先对高分辨率的遥感影像进行预处理，包括遥感影像的大气校正与研究区域裁剪，对处理后的各波段进行波段运算以提取先验特征信息；应用多源数据融合实现多波段和特征信息的融合，构建数据集；训练并验证由卷积神经网络搭建的语义分类模型ME‑net；调用ME‑net模型实现红树林的自动分类，输出一个png格式的掩模文件，即为分类和提取的结果；通过长距离条件随机场对分类结果进行细调。本发明中的分类模型在应用中可通过扩展数据集使得分类精度达到92.3％，完全可以代替人工目视解译，为高精度影像地图的更新和滨海地区生态系统的保护提供辅助技术支持。 【EN】The invention discloses a remote sensing image mangrove forest extraction method and system based on a deep convolutional neural network, firstly preprocessing a remote sensing image with high resolution, including atmospheric correction and research area cutting of the remote sensing image, and carrying out wave band operation on each processed wave band to extract prior characteristic information; fusion of multiband and characteristic information is realized by applying multi-source data fusion, and a data set is constructed; training and verifying a semantic classification model ME-net built by a convolutional neural network; calling an ME-net model to realize automatic classification of mangroves, and outputting a png-format mask file which is a classification and extraction result; and fine-tuning the classification result through a long-distance conditional random field. The classification model can ensure that the classification precision reaches 92.3 percent by expanding the data set in application, can completely replace manual visual interpretation, and provides auxiliary technical support for updating a high-precision image map and protecting an ecosystem of a coastal region.

摘要：【中文】本发明公开了一种高性能的车辆动态三维建模与轨迹实时渲染方法及装置，先截取视频画面为图片，记录每张图片中车辆的屏幕坐标和车辆类型，并存储在txt文件中；预先设置好的道路模型以及车辆模型导入到MapGIS数据库中；在MapGIS桌面软件里构建三维场景，添加导入后的道路模型，并保存为地图文档；手动采集视频截图和三维场景中的四个控制点对；调用桌面二次开发接口在场景中预先添加不同类型的车辆模型，不设置模型的位置和角度信息，显示状态设为不可见；设置一定时间间隔顺序读取txt文件，根据坐标转换方法计算出车辆的三维坐标，根据车头车尾坐标，计算出车辆的角度，动态更新场景中预先添加的小车坐标、行驶角度，并设置模型的显示状态设为可见。 【EN】The invention discloses a high-performance vehicle dynamic three-dimensional modeling and track real-time rendering method and a device, wherein a video picture is firstly captured as a picture, the screen coordinate and the vehicle type of a vehicle in each picture are recorded and stored in a txt file; importing a preset road model and a preset vehicle model into a MapGIS database; constructing a three-dimensional scene in the MapGIS desktop software, adding the imported road model, and storing the road model as a map document; manually collecting four control point pairs in a video screenshot and a three-dimensional scene; calling a desktop secondary development interface to add different types of vehicle models in a scene in advance, setting no position and angle information of the models, and setting the display state to be invisible; setting a certain time interval to sequentially read the txt file, calculating the three-dimensional coordinates of the vehicle according to a coordinate conversion method, calculating the angle of the vehicle according to the coordinates of the head and the tail of the vehicle, dynamically updating the coordinates and the running angle of the trolley added in advance in the scene, and setting the display state of the model to be visible.