The RID Rapid Automatic Detection Method In Freeways

With the development of transportation industry,the ceaseless construction of highway and the increase number of mileage,at present,the total number of highway mileage has been the NO.1 around the world.It is coming soon of the large-scale maintenance in highway.So the Pavement Management System(PMS)is introduced to accomplish the task of pavement-maintenance.Pavement diseases can be recorded and located by Linear Mileage Location in PMS.However,there are kinds of problems in the Linear Mileage Location,which is the management units of pavement being divided too large.this problem caused unfining maintenance management and inaccurate location.The RID theory is proposed to solve the problem of inaccurate location and broken chainages.The RID is represented the smallest management units of pavement.This theory can efficiently solve the problem of inaccurate location.But,it is a challenge task to collect and recognize the RID marks which are characterized by large-distribution-density,long mileage and small area.It is still an empty field in the research of auto-collecting and recognizing the RID mark at present.This paper proposed a rapid and efficient collection plan,aimed at RID,and a location method based on a double-threshold.This method is using high and low threshold to process the same RID pic at the same time,and then add the two processed pics,and get the wanted pic at last.The key part in this method is the threshold of high and low.This paper gets the high and low threshold by seeking the local extremum depending on the curve of image itself gray value.It is tested by experiments that the precision rate of RID location based on double-threshold method is higher than the one based on the traditional houghline test.Finally,this paper uses BP neural network to recognize the characters of RID,which could efficiently recognize the incomplete characters.This paper selects 1000 RID images to locate and recognize,which location precision rate reached 95%,the recognition precision rate reached 92.5%.It is tested by experiments that the location method in this paper is highly efficient and feasible,and this system runs safe and reliable.