automated-road-extraction

Automated roads extraction from satellite imagery using a combination of image processing techniques and dynamic path finding algorithms. Read the full project report here

Getting Started

These instructions will get you a copy of the project up and running on your local machine for use. All sample images are present in data directory.
Read the description of each step here

REQUIREMENTS

1. Python 3.x
2. Python Libraries:
   • OpenCV-Python 3.1.0
     
   • numpy
     
   • scikit-image

Inputs

• Image name (python run.py --imagename)
• Input image number after completion of step 2. Step 3 requires manual selecting best image from results of step2 variance test for further processing.

Running the Code

To run the code, use the following command in the main directory

• python run.py --image_name

Example Run

python run.py t3

Processing Image t3

       STEP 1: Applying Morphology          

Kernel Size: 25x25
Applying open close operations
Saved OTSU_thresh.png, noise.png, noise_free.png in the directory " Morphology\t3 "

| | |

    STEP 2: Applying Variance Test         

Thread 0 processing Patch Length: 30 Patch Width: 5
Thread 1 processing Patch Length: 30 Patch Width: 10
Thread 2 processing Patch Length: 40 Patch Width: 5
Thread 3 processing Patch Length: 40 Patch Width: 10
Variance_test\t3\30_5_2.png Saved
Variance_test\t3\30_10_2.png Saved
Variance_test\t3\40_5_2.png Saved
Variance_test\t3\40_10_2.png Saved

| | |

Select image from " variance_test\t3 " for further processing
1 -> 30_5_2.png
2 -> 30_10_2.png
3 -> 40_5_2.png
4 -> 40_10_2.png
Enter image number: 1

  STEP 3(a): Single Pixel Path Detector     

Single Line Thread 0 Length Thresh: 50 Angle Thresh: 3
Wide Jump Thread 0 Length Thresh: 60 Angle Thresh: 6
Single Line Thread 1 Length Thresh: 50 Angle Thresh: 6
...

   STEP 3(b): Wide Jump Path Detector      

...
wide_jump_path\t3\Vertical Paths\80_12.png Saved
wide_jump_path\t3\Horizontal Paths\80_12.png Saved
single_pixel_path\t3\Vertical Paths\70_3.png Saved
single_pixel_path\t3\Horizontal Paths\70_3.png Saved

| | |

         STEP 4: Path Refiner                

-- Kernel Size: 5x4

Refining Wide Jump Path Images
Refining Single Pixel Path Images

Projecting Refined Images onto the Satellite Image

Processing Complete

All resulting images saved in Results directory

| | |

Built With

• python