Research On Application Of Image Recognition In Yard Process Control

Modernized railway and railway intelligent systematic development, put forward high requirement in security, dependability, real-time character, respect such as being comprehensive marshaling yard. Subject this to how research adopt advanced image recognition technology in marshalling yard hump application, homework of course just. Utilize the pattern recognition theory, direct against the hump homework situation in the marshalling yard, put forward one through picture array fast, intuitive judgment is it set car group characteristic free and walk method , conduct of trend to slip away, improve the precision of measuring, improve the existing deficiency of measuring the method . The static picture shot to the hump field is indicated by the pattern process and realization of the algorithm, adopt the proper treatment method, can get the better experimental result. The main research contents are as follows:(1) Analyse the current situation of domestic and international hump development of marshalling yard, increase year by year to the volume of goods transported of marshalling yard at present of our country, homework security demand actual conditions that raise constantly , combine the advanced scientific and technical result, put forward the view of adopting application in the hump homework course of marshalling yard is controlled of recognition technology of picture .(2) To realize hump stand theoretical foundation and some methods , field of pattern recognition carry on the discussion. Focus on pattern process , characteristic and draw the application that measured the respect with the movement. Pattern recognition means mainly that adopts the digital technology method through the computer, the picture obtained to a systematic front carries on corresponding treatment according to the specific purpose . Usually the recognition system of the mode is made up of 4 parts: Obtain , pretreatment , characteristic drawing and choosing , categorised decision in data. Have carried on deeper researching and analysing to relevant theories. And shoot the picture on the hump field , through the embedding in pattern process of Mat lab , have got the static picture which passes the pretreatment of the picture . Experiment indicate adopt the proper treatment method, can get the better experimental result.(3) In order to make a video recording the picture is as the target , proceed from treatment technology of the picture , carry on deeper research , is accused of the target's main mode in the hump to the marshalling yard , regard slipping away and showing the car group of the homework as target, walk quotations condition , is it put speed differentiate to slip away to various kinds of mode car group, it establish calculate by distance with actually from relation model. Field head apply the brake to purpose interval location put camera , camera shoot with certain time interval car group slip away state of setting free in hump, input computer picture that shoot, set characteristic , car of group free and walk through pattern process and pattern-recognition method conduct trend judge slipping away, can make hump homework control result to be good.(4) In the course of discerning, on the basis of using traditional pattern-recognition theory, to some new methods, new technology carries on some tries, has put forward some one's own new opinions, also made a certain analysis and discussed about the existing problem as regards concrete conditions of what a scheme facesat the same time , because marshalling yard rail position does not change, is it set free car group move , and just along regular direction in two-dimentional translation of level sport its to slip away only, picture that obtain each time except picture that cause by weather reason grey level to some extent getting different, a only one difference to is it set car change , group of position free to slip away. And so long as we discern some some situations of going out with the car in the group, can find out about the situation of the whole car group . Because the style of the truck is confirmed basically, the priori information of solving the pattern-recognition is very sufficient , make pattern process and method that discern simplify greatly like this.