CCTV footage has improved in leaps and bounds as we made the transition from analogue to digital with hard drives replacing tapes. Despite this, video surveillance footage is still vast and it is laborious to "sift" through images in order to find evidence. This requires some form of intelligence that can extract useful data from video footage automatically. Enter video analytics.
Although video analytics might sound simplistic, it has taken years of development and many technology iterations to get to the point of maturity that we have reached today. Fortunately, there are a few applications in the transportation technology that can benefit distinctly from this technology.
The most common application is Automatic Number Plate Recognition (ANPR), which makes use of Optical Character Recognition (OCR). The number is "read" a few times on the same vehicle and if the same number is recognised or found, this increases the accuracy of the reading. However, the faster the cars are travelling, the less accurate the video analytics. It must be noted that there is still technology refinement in the pipeline as many systems must view a number plate directly from the front and not the side, which can result in misreads due to distortion. Further exacerbating this are the provincial images printed behind the number and the fact that number plates do not incorporate an international character set. Added to this, vehicle lights "blind" the camera and require infrared lighting and lens filters to eliminate this problem.
A more up-to-date example of video analytics is with our toll plazas that are currently being completed. Video analytics is used to manage the risk of fraudulent transactions, working in tandem with a video "grabbing" system that monitors the vehicles going through the toll plazas. Future developments could include monitoring speed by calculating the time from one gantry to the next.
As with many systems, what would truly add value is integration of information obtained through video analytics with other systems such as eNatis and even the Department of Justice. Consider the possibilities, such as criminals on the loose being tracked by linking their identity numbers to their vehicle registration numbers and possibly even other cases that are outstanding. The benefits are endless.
The answer, in large part, lies in identifying a solution provider that has the product and industry knowledge, as well as the experience, to make the technology meet the unique needs of the transportation industry. Having a partner that understands and can deliver the "end to end" solution, plus integration capabilities, provides the benefit of fewer challenges with implementation and maximising the solution by leveraging other systems and applications. The other part of the puzzle is in understanding the limitations of video analytics.
Sophisticated software algorithms analyse incoming video streams to enable licence plate recognition, facial recognition and behavioural recognition, along with advanced archiving and retrieval capabilities.
In a busy airport, however, a video analytics solution is unlikely to identify an abandoned suitcase or detect a change in crowd behaviour - it's just too busy an environment. And, regardless of the environment you install it in, the solution is very likely to need ongoing fine-tuning, including factoring in recognition of seasonal environmental changes.
Although there are some challenges still to be resolved with video analytics, this technology is enhancing the transportation sectors ability to monitor and prevent fraud, a key concern for South African businesses that are positioned in this area. Choosing a partner that has the experience, skills and expertise in this area will greatly enhance the functionality and return of investment.
