The Discernment And Measuring Of The Automobile License Plate

This text mainly researches automobile license plate automatic recognition system character cutting and algorithm, through a large number of research of target compare,in every deal with module provide practical and pointed algorithm, concrete innovation as follows: In automobile license plate picture unity, background of color, utilizing four separate automobile license plate picture space agree with message that the background distribution . 1. Some grey level value the background under in character ; 2.Control character alternate place background some grey level value (utilize vertical operator , Sobel of direction , neglect competence interfere, narrow the hunting zone); 3. The shape distribution: Utilize black bottom mispronounced character and white bottom black word picture correctly in the morphology thick to take treatment appear the characteristic of the antithesis ----Character expand or shrink, make to treated picture vertical projection find out two pictures with characteristic that luminance distribution; 4. Effective area ratio utilize character picture element With the character of area preponderating the characteristic, put forward effective concept of area , there are empty backgrounds that are filled character area, neglecting the row that must be the background area in order to strengthen the dependability of this parameter at the same time. In the two accurate value, utilizing in histograms big young path breaker's picture of peak value is classified near expected value, making different threshold value search. Have put forward nearby expected in the histogram and searched for two best value threshold value: Limit the hunting zone and hope to deduct it by a definite date between a abundant quantity and histogram spike, searching, weighing out the best threshold value from the upper and lower limits of this area separately. Using horizontal bright dark change as according to go except about frames interfere; Except separate incantations by surveying white some methods of position of picture element; Going with marking law (two value picture person who border on ) except that left side interfere. In the treatment cutting apart in the character, utilizing vertical gray level projection as the rough basis that cuts apart, then comprehensive character one's own width information and unsmooth change, making meticulous cutting apart further of the character of shape.In a word, basing on the premise that the big module of system is setting, this text carefully researches the characteristics of target , making the concrete work of treatment have an object in view, the validity and terseness of the algorithm given consideration , meeting the treatment demand .Of course, because of the restriction on personal ability and time, the algorithm itself still has some mistakes, there is certain one that is deficient in considering the comprehension of the question. Discussing as follows: Two value methods the text adopts is still a kind of overall threshold value law after all , though attempting to break through the limitation as threshold value directly, adaptation all license plates on but not perfect. And this is the place where overall threshold value is impassable, because we have not considered different characteristics in different areas of whole license plate, making the adaptability of the algorithm type less better. We should see that, as regarding this research object of license plate of the automobile, two value methods has more improvement . For example, if can...