Sei ISOBE

Research on Development of the Road Surface Freeze Detection System Using Image Processing

Kunio TORII

In a snowfall area, it is important to secure the traffic safety of winter and the safety of an economic activity.Moreover, when people lead a life, the information on a road surface situation is indispensable. Therefore,the road surface freeze detection system is developed as the snow coverage of a road surface, the situation of a freeze, etc. are judged,and a system which provides people with information now. This system collects data required in order to judge the state of a road surface by information gathering apparatus, such as a sensor.Based on this, it judges whether the state of a road surface is dry, or wet.Next, outside temperature and road surface temperature are measured . Finally a road surface situation is judged together with these. And people are provided with the obtained judgment result as information through radio or a lightning sign.

Now, an information gathering apparatus has an optical fiber, a static electricity capacity , etc. As the sensor of optical fiber is a line of detection , another is a point ( a small domain about 80cm2 ) , so neither covers the road surface. Moreover, each distinction of water, ice, and snow cannot do them. Furthermore, the sensor of laying-under-the-ground formulae produce an incorrect operation of friction, degradation, etc. A problem is in these detector accuracy and reliability.

Therefore, in this research, it experimented in an inside-of-a-house experiment or the road surface freeze spot (Omachi-shi, Nagano), and resulted in construction of the road surface freeze detection system by image processing using the infrared camera and the CCD camera as a method detectable non-contact and broadly .
The road surface freeze detection system built by this research is as follows.Between after a road surface freezes until it melts was divided by two STEP. STEP‡@ is the case where the dryness road surface is a wet road surface by rain or snow. Here, using an infrared camera presumes road surface temperature. If 5 degrees C or less, it moves to STEP‡A. STEP‡A distinguishes road surface situations (a freeze, snow coverage, etc.) and a freeze portion using a CCD camera, is observed until they stop having melted and it becomes a wet road surface. The processing method is background subtraction. The PIXEL number more than the threshold of created 2 value-ized picture is counted, and time differentiation of it is carried out. And when a background picture is made into a wet road surface, the maximum of differentiation begins to carry out a road surface freeze.While when a background picture is made into a wet road surface,the minimum of differentiation becomes a wet road surface. So if judged as a wet road surface, it will move to STEP‡@ and road surface temperature will be presumed. Furthermore, when the data which counted the PIXEL number was applied to the growth curve, it turns out that it is possible to predict a road surface freeze.