Man has been developing various methods to protect himself from natural calamities since ages. The only scientific solution to natural calamities is development of systems to predict, detect and take preventive measures using recent advancement in technology. Along the highly landslide prone Konkan railway line, many people have lost their lives due to landslides. It is now high time to replace the present obsolete manual detection systems deployed along this line. In this paper, a highly accurate, effective and efficient landslide detection system has been proposed which can be used along the Konkan railway line to monitor tracks for landslide using image processing. The coding has been done using MATLAB and a low resolution webcam was used for acquiring sample video frames. Various techniques like Hamming distance, Entropy, Euclidean Distance, Correlation, Block processing etc. were used for event detection. The proposed technique gave a threshold margin of 80.24% and the average efficiency of the system was found to be 86.67% for the considered set of images. Using proposed technique, False Acceptance Ratio (FAR) of 0.067 and False Rejection Ratio (FRR) of 0.933 were achieved.

No abstract