Railway Track Fault Detection Project

Download Document/Synopsis
Here we propose an innovative approach to detect railway track crack as this system detects crack based on image processing. Many image preprocessing steps is used to detect railway track crack. As image is prone to noise. System converts image to grayscale image and uses filtering to remove noise from image. Noise removal helps to detect crack more accurately. Image luminous level is increased and image is converted to binary image. This helps system to detect only crack and helps to remove other unwanted objects. Image once converted to binary image, holes are filled by using image processing method this helps to reject all smaller objects which are not required for crack detection. Intensity value is used for accuracy purpose. Blob analysis method is used to detect large blobs. System detects crack based on number of connected components. System detects crack based on number of blobs involved and mentions whether crack exist or not. Using bounding box functionality, system displays rectangular box around the blob. This system used during railway track inspection. The proposed system is able to detect deeper cracks with 80% success rate and minor cracks with 50-60% accuracy


Advantages
  • System will help to reduce accidents caused due to railway cracks.
  • This system helps track inspection coach to complete track inspection faster.
  • No Manual intervention needed.
Disadvantages
  • User must provide good image. System will not work properly with damaged image.
  • Low Accuracy on improper lighting.

-->