C#ArcEngine二次开发影像去黑边问题

C#ArcEngine二次开发影像去黑边问题
我的代码如下，通过设置nodata值进行去黑边，

                 ESRI.ArcGIS.DataManagementTools.CopyRaster rc = new ESRI.ArcGIS.DataManagementTools.CopyRaster();
                                rc.in_raster = fName;//
                                rc.out_rasterdataset = tifname;
                                rc.background_value = 0;
                                rc.nodata_value = "0";
                                rc.pixel_type = "32_BIT_FLOAT";
                                ESRI.ArcGIS.Geoprocessor.Geoprocessor GP = new ESRI.ArcGIS.Geoprocessor.Geoprocessor();
                                GP.Execute(rc, null);
        }

但是效率很低，处理时间在5分钟以上，因为影像也比较大。请问有没有其他方法提高效率进行去黑边处理

引用chatGPT作答，您可以尝试使用基于像素值的遮罩来去除黑边。以下是一些建议：

1.计算图像的边界像元。您可以使用Con工具来实现这一点。例如，您可以将像素值为0的像元（黑色）设置为1，而其他像元（非黑色）设置为0。然后，您可以使用RasterToPolygon工具将该栅格转换为多边形要素。

2.缓冲区多边形要素。这将为每个多边形创建一个缓冲区，以便在稍后使用。缓冲区的大小应根据影像的黑边大小来设置。

3.将缓冲区要素转换为遮罩。使用FeatureToRaster工具将缓冲区要素转换为栅格遮罩。

4.将原始影像和栅格遮罩相乘。使用Times工具将原始影像和栅格遮罩相乘。这将移除黑边并创建一个新的栅格数据集。

以下是一个基于上述步骤的C#代码示例：

ESRI.ArcGIS.DataManagementTools.Con con = new ESRI.ArcGIS.DataManagementTools.Con();
con.in_conditional_raster = fName;
con.in_true_raster_or_constant = 0;
con.in_false_raster_or_constant = 1;
con.out_raster = "boundary.tif";
ESRI.ArcGIS.Geoprocessor.Geoprocessor GP = new ESRI.ArcGIS.Geoprocessor.Geoprocessor();
GP.Execute(con, null);

ESRI.ArcGIS.DataManagementTools.RasterToPolygon toPoly = new ESRI.ArcGIS.DataManagementTools.RasterToPolygon();
toPoly.in_raster = "boundary.tif";
toPoly.out_polygon_features = "boundary.shp";
toPoly.simplify = "NO_SIMPLIFY";
GP.Execute(toPoly, null);

ESRI.ArcGIS.AnalysisTools.Buffer buffer = new ESRI.ArcGIS.AnalysisTools.Buffer();
buffer.in_features = "boundary.shp";
buffer.out_feature_class = "buffer.shp";
buffer.buffer_distance_or_field = "10"; // 缓冲区大小根据黑边大小调整
buffer.line_side = "FULL";
buffer.line_end_type = "ROUND";
buffer.dissolve_option = "NONE";
GP.Execute(buffer, null);

ESRI.ArcGIS.ConversionTools.FeatureToRaster toRaster = new ESRI.ArcGIS.ConversionTools.FeatureToRaster();
toRaster.in_features = "buffer.shp";
toRaster.out_raster = "mask.tif";
toRaster.field = "GRIDCODE";
toRaster.cell_size = fName;
toRaster.setnull_values = "0";
GP.Execute(toRaster, null);

ESRI.ArcGIS.SpatialAnalystTools.Times times = new ESRI.ArcGIS.SpatialAnalystTools.Times();
times.in_raster_or_constant1 = fName;
times.in_raster_or_constant2 = "mask.tif";
times.out_raster = tifname;
GP.Execute(times, null);

以下回答参考GPT，并由JoseKe整理完成，希望您能采纳：
可以使用多线程的方式对大型数据集进行处理来提高效率。同时，可以考虑以下优化：

1.使用栅格数据处理器而不是地理处理器来处理栅格数据，栅格数据处理器比地理处理器更快；

2.将数据集进行分块处理，将大型数据集分为多个小型数据集并行处理，使每块数据可以在单独的线程中进行处理；

3.使用图像处理算法来处理黑边问题，比如边缘检测算法或者其他的形态学运算来处理黑边问题，可以提高处理效率。

下面是使用多线程处理数据集的代码示例：

ESRI.ArcGIS.DataManagementTools.CopyRaster rc = new ESRI.ArcGIS.DataManagementTools.CopyRaster();
rc.in_raster = fName;// 
rc.out_rasterdataset = tifname;
rc.background_value = 0;
rc.nodata_value = "0";
rc.pixel_type = "32_BIT_FLOAT";
ESRI.ArcGIS.Geoprocessor.Geoprocessor GP = new ESRI.ArcGIS.Geoprocessor.Geoprocessor();

//获取数据集范围
IRasterDataset2 rds = GP.Open(raster_path) as IRasterDataset2;
IRasterProps rProps = rds as IRasterProps;
IPnt tileSize = new PntClass();
tileSize.SetCoords(256, 256);
IGPValue TileSize = new GPValueClass();
TileSize.SetAsText(tileSize.X.ToString() + " " + tileSize.Y.ToString());
IRasterDataset3 rasterDs = rds as IRasterDataset3;
IRasterTypeProperties rasterTypeProps = rasterDs as IRasterTypeProperties;
IGPUtilities gpUtils = new GPUtilitiesClass();
List<string> BlockPaths = new List<string>();
int MinBlockNumber = gpUtils.GetCount(rds, tileSize)[0];
rasterTypeProps.MeanCellSizeProperties = rProps.MeanCellSize();
int tiles = 0;
while (tiles < MinBlockNumber)
{
     try
        {
        string block = Path.GetDirectoryName(tifpath) + "\\" +Path.GetFileNameWithoutExtension(tifpath) + "_block_" + tiles.ToString() + ".tif";
        ESRI.ArcGIS.DataManagementTools.ExtractSubDataset extractSubDataset = new ESRI.ArcGIS.DataManagementTools.ExtractSubDataset();
        extractSubDataset.in_raster = raster_path;
        extractSubDataset.out_raster = block;
        extractSubDataset.where_clause = "OBJECTID >= " + ((tiles) * tileSize.X * tileSize.Y + 1) + " AND OBJECTID <= " + (((tiles + 1) * tileSize.X * tileSize.Y) >= rProps.Width * rProps.Height ? rProps.Width * rProps.Height : ((tiles + 1) * tileSize.X * tileSize.Y));
        GP.Execute(extractSubDataset, null);
        BlockPaths.Add(block);
        }
    catch(Exception e)
        {
        MessageBox.Show(e.Message);
        }
    tiles++;
}

//新建多线程处理块数据
int maxThreads = 8;
int blockCount = BlockPaths.Count;
Semaphore sem = new Semaphore(maxThreads, maxThreads);
List<Thread> threads = new List<Thread>();
for (int i = 0; i < blockCount; i++)
{
     sem.WaitOne();
     Thread t = new Thread(new ParameterizedThreadStart(Execute));
     t.Start(new object[] { BlockPaths[i], rc });
     threads.Add(t);
}
//等待所有线程执行完毕
foreach (Thread t in threads)
{
     t.Join();
}

其中Execute函数为：

private void Execute(object o)
{
     object[] args = o as object[];
     string raster = args[0].ToString();
     ESRI.ArcGIS.DataManagementTools.CopyRaster rc = args[1] as ESRI.ArcGIS.DataManagementTools.CopyRaster;

     ESRI.ArcGIS.Geoprocessor.Geoprocessor GP = new ESRI.ArcGIS.Geoprocessor.Geoprocessor();
     GP.Execute(rc, null);

     sem.Release();
}