Initiatives at IIT Delhi and BETQ Data Analytics Pvt Ltd, India
At IIT Delhi, as a part of doctoral research, we have developed an improved method for automated detection and analysis of road surface distresses using video image based techniques. The developed method is performed in two stages. In the first stage, the database of video clips captured with/without distress is processed with the developed ‘distress frames selection (DFS)’ algorithm for rapid screening of video frames that have a high probability of containing distress from those that have a very low probability. This algorithm also gives an assessment of general well being of the road. Further, it increases the accuracy of distress identification in the second stage of processing. The DFS algorithm categorizes video frames into ‘frames without distress’ and ‘frames with distress’. These frames are further tagged with their respective frame numbers or locations. In the second stage of processing, the video frames belonging to ‘frames with distress’ are further analyzed with the developed critical distress detection, measurement and classification (CDDMC) algorithm. The CDDMC algorithm provides simultaneous detection, measurement and classification of road distresses in one pass. The result is categorization of frames as ‘frames with cracks’, ‘frames with potholes’, ‘frames with patches’ and ‘frames without critical distress’. Besides, frames are also tagged with the type of distress identified such as potholes, cracks and/ or patches. Measurement of distress information in the frame is reported in a printable format.
Economic automated Solutions for Indian conditions
In the context of road inventory and condition survey, the need of the hour is the automation in processing/analysis of the collected road video/image data for useful information extraction. In this dimension, several research efforts have been made in the past around the globe for development of automated solutions using various approaches and techniques. As a result, many products could be developed today. Examples of such systems employing automated processing methods are Automatic Road Analyzer (FugroRoadware, Canada), RoadCrack System (Commonwealth Scientific and Industrial Research Organization, Australia) and Digital Highway Data Vehicle (Dynatest and WayLink, USA). But, these systems were used for automated cracks detection and analysis only and they were developed based on the local conditions prevailing in the developed countries like USA, Australia and Canada etc.
For Indian road conditions, BETQ, a research driven Start-up Company born in India, has developed various automated software products such as AutoDistress, AutoPotholes, AutoCracks and AutoPatches. These software products can detect and measure potholes, cracks and patches automatically with 90% accuracy from 2D-video clips collected by road survey vehicles using low-cost / high-cost video cameras. However, the reliability of distress identification and the accuracy of distress quantification could be achieved 100% at the current time due to various challenges.
The key challenges before researchers in developing reliable solutions for automation in road inventory and condition survey are automated segmentation and classification of road assets and surface distresses out of video clips of Indian roads, collected in real life conditions with different images of objects (vehicles parts, man, animals and manholes), different shadows of various objects (vehicles parts, man, trees, animals and poles) and different road markings (white, yellow, and black) that may occur perchance. Besides, there are many classifications / parameters of road condition data such as potholes area & number, cracks area, patches area, ravelling area, rutting area & depth, bleeding area, edge break area, corrugated area, shoulder condition and side drain condition etc. There are as many road asset inventory parameters such Road Class Type, measurements of Right of Way, Formation Width, Width of Carriageway, Width of Shoulder, Average Height of Embankment, Type of Pavement, Type of Carriageway and Type of Terrain. This makes the automation process more complicated and costly. As such a little amount of human intervention is essential for higher accuracy and reliability in processing of the raw video data collected in real life conditions. Thus, semi-automated road video data processing software products like SemiDistress and SemiAsset from BETQ, India are found to be more suitable for Indian conditions at the current time. These software products can also give classification and measurement of road surface distresses and assets /events with higher accuracy than that of manual / automated methods of video data processing.
One of the economical road survey systems which is readily available in India at very competitive price is the PARSS (Portable Automated Road Survey System) from BETQ.
PARSS has been developed through doctoral level research and it has been designed to be very portable, easily detachable, plug & play / recording type, modular system with its automated / semi-automated data processing software. At present, it includes two video Systems Pavement View Video System and Asset View Video System.