【SuperMap iDesktopX 2024视频分析】

---

前言

  SuperMap iDesktopX 2024基于AI与AR技术，提供的视频分析功能，使得我们具备更强的视频感知能力，因而从视频中提取更为丰富的信息。视频分析模块提供了目标检测、车牌识别、围栏分析、轨迹提取，以及AR空间分析功能，包括缓冲区分析、空间查询、空气质量分析等，可将视频与GIS结合，在视频地图中提取更多的空间信息，并基于视频进行空间分析。以下是相关环境配置以及目标检测、地理围栏的示例代码。

---

一、视频分析环境配置

1.AI环境配置

  在使用视频分析功能之前，需要先下载 SuperMap iDesktopX 的AI扩展包，并进行相关的环境配置，配置好环境后即可激活视频分析中的功能。具体环境配置可以参考：https://help.supermap.com/iDesktopX/zh/SpecialFeatures/ARMap/ARAnalysis/AREnvironmentConfiguration.html

2.开发环境配置

  在iDea"Project Structure - Project Setting - Libraries"中引入以下依赖：
●%iDesktopX产品包%/bin/
●%iDesktopX产品包%/bundles/idesktop_bundles/
●%iDesktopX产品包%/bundles/require_bundles/
●%iDesktopX产品包%/lib/

二、视频分析扩展代码

1.加载视频分析环境

代码如下：

//加载OpenCV，用来解析视频
		LoadDllManager.loadDlls();
		//加载Python环境
		PythonUtilities.setPythonExecutor(new PythonDefaultExecutor());
		GlobalParameters.setHighPerformanceDetection(true);
		GlobalParameters.setSupportStartPythonProcess(false);
		VideoAnalystRegister.getInstance().register();

		//加载视频分析环境
		VideoPlayerPython videoPlayerPython = new VideoPlayerPython();
		videoPlayerPython.getVideoAnalystEnvironment();

		//打开空间化后的视频数据集
		Workspace workspace = new Workspace();
		String udbxPath = "F:\\Temp\\Video9SingleDatasetMultiVideo.udbx";
		DatasourceConnectionInfo datasourceConnectionInfo = new DatasourceConnectionInfo();
		datasourceConnectionInfo.setServer(udbxPath);
		datasourceConnectionInfo.setEngineType(EngineType.UDBX);
		datasourceConnectionInfo.setAlias(FileUtilities.getFileNameWithoutExtension(udbxPath));
		Datasource datasource = workspace.getDatasources().open(datasourceConnectionInfo);
		DatasetVector datasetVector = (DatasetVector) datasource.getDatasets().get("video9_1Video");

		//遍历视频数据集并提交为视频分析任务
		Recordset recordset = datasetVector.getRecordset(false,CursorType.STATIC);
		while (!recordset.isEOF()){
			VideoManager videoManager = new VideoManager(recordset);
			videoManager.build();
			recordset.moveNext();
			VideoAnalystExecuteEngine.getInstance().addVideoTask(new VideoTask(videoManager));
		}
		RecordsetUtilities.releaseRecordset(recordset);

2.目标检测

代码如下：

/*
		 * 目标检测代码
		 * */
		//开启目标检测
		videoPlayerPython.getVideoAnalystEnvironment().startAnalyst(VideoAnalystFunctionKey.OBJECTION_DETECTION);
		VideoAnalystObjectDetection videoAnalystObjectDetection = (VideoAnalystObjectDetection) videoPlayerPython.getVideoAnalystEnvironment().getVideoAnalyst(VideoAnalystFunctionKey.OBJECTION_DETECTION);

		//设置房屋模型
		AnalystModel analystModel = AnalystModelManager.getInstance().getAnalystModel("TieTaFangWu");
		videoAnalystObjectDetection.getVideoAnalystParameter().setAnalystModel(analystModel);
		videoAnalystObjectDetection.getVideoAnalystParameter().getSupportObjectDetection().clear();

		//设置目标检测类型和结果绘制风格
		for (DetectionType detectionType : analystModel.getDetectionTypes()){
			videoAnalystObjectDetection.getVideoAnalystParameter().getSupportObjectDetection().put(detectionType.getStrType(), VideoAnalystUtilities.getObjectDetectionStyleByBackgroundColor(Color.GREEN));
		}

3.地理围栏分析

代码如下：

/*
		 * 地理围栏分析代码
		 * */
		//开启围栏分析
		videoPlayerPython.getVideoAnalystEnvironment().startAnalyst(VideoAnalystFunctionKey.VIOLATION);

		//获取围栏分析实例
		VideoAnalystViolation videoAnalystViolation = (VideoAnalystViolation) videoPlayerPython.getVideoAnalystEnvironment().getVideoAnalyst(VideoAnalystFunctionKey.VIOLATION);
		VideoAnalystViolationParameter videoAnalystViolationParameter = videoAnalystViolation.getVideoAnalystParameter();

		//设置取证存过滤错误结果
		videoAnalystViolationParameter.setViolationIntersectAreaRatio(0.5);
		videoAnalystViolationParameter.setViolationOutputFolderPath("F:\\SuperMap\\result");
		videoAnalystViolationParameter.setOutputImageTimeInterval(1);

		//设置围栏分析规则
		DatasetVector selectedDataset = (DatasetVector) datasource.getDatasets().get("GengDi");
		videoAnalystViolationParameter.setViolationDatasetVector(selectedDataset);
		Geometry geometryByViolationExpression = VideoAnalystUtilities.getGeometryByViolationExpression(selectedDataset,"SmID>0");

		SpacialAnalystData spacialAnalystData = new SpacialAnalystData("耕地违建","SmID>0",geometryByViolationExpression,true);
		spacialAnalystData.setViolationJudgmentRule(ViolationJudgmentRule.INCLUDED);
		videoAnalystViolationParameter.getViolationData().add(spacialAnalystData);

4.获取视频分析结果

代码如下：

/*
		 * 获取视频分析结果
		 * */

		//添加获取视频分析结果事件
		videoPlayerPython.getVideoAnalystEnvironment().addVideoAnalystResultChangedListener(new VideoAnalystResultChangedListener() {
			@Override
			public void videoAnalystResultChanged(VideoAnalystResultChangedEvent videoAnalystResultChangedEvent) {
				//获取视频分析结果
				ArrayList<VideoAnalystResult> currentVideoAnalystResults = videoPlayerPython.getVideoAnalystEnvironment().getCurrentVideoAnalystResults();
				for (VideoAnalystResult currentVideoAnalystResult:currentVideoAnalystResults) {
					pushVideoAnalyst(currentVideoAnalystResult);
				}
			}
		});

		//执行分析任务
		VideoAnalystExecuteEngine.getInstance().addVideoPlayerPython(videoPlayerPython);
		VideoAnalystExecuteEngine.getInstance().run();

三、代码运行结果

  完整工程代码以及视频下载链接：https://pan.baidu.com/s/1-8VOwKPWhMap6g_1HscdXw?pwd=8e64  运行工程请参考SuperMap iDesktopX二次开发项目搭建流程中1.2小节。

四、桌面功能

  除了扩展代码外、SuperMap iDesktopX 2024桌面端也有视频检测功能，参考1.1AI环境配置，配置完成之后，打开SuperMap iDesktopX 2024桌面端，具体操作步骤可以查看帮助文档中目标检测。以下是桌面端运行结果：