Tool: ArcGIS Pro Desktop 2.1
Original data: 9 grayscale images in Shanghai area

Combine image parts

Using Mosaic to stich images into onw big image (click to see clearer images).

Parameters	Result

Convert grayscale image to binary

In grayscale image, the value of each pixel can range from e.g. [0-255]. But the bianry image pixel value can be either 0 or 1.

This step is to set a threshold, which will be used in classify grascale pixel values into 0 or 1.

The threshhold is vital in the whole processing, e.g. some grascale images can have blurring building edges. This threshold determines the edge shape in vector images. The threshold value can be different in every cases.

Reclassify function is used in this step.

Parameters	Result

Automatic vectorization

Using Raster to Polygon (or Raster to Polyline) to vectorize the binary image. P.S. If the image size is big, this step can take a long time.

Parameters	Result

Here we have the voctor image. But if you check the image closely, you will find some small islands on the image and some jagged edges. In order to remove them, the image needs to be furture processed.

Projection

If the image has a Geographic Coordinate System, we need to project it into a Projected Coordinate System.

Using Project to reprojection the image.

Parameters	Result

Simplification

Using Simplify Polygon to remove the small islands and smoothify the edges.

Parameters	Result