| License plate location and character segmentation in complex environment has always been the focus and difficult point of license plate recognition technology.Many scenes in complex environment seriously affect the speed and accuracy of license plate location and character segmentation,leading to the failure of license plate recognition finally.Such as the existence of reflective,backlight,uniform illumination;uncertain shooting angles and varying shooting distances;license plate texture features are similar or there are car lights,bumpers,license plate signs and advertising information plates similar to the license plate background color;there are defilement,character adhesion and fracture of the license plate.Aiming at the difficulties of license plate location and character segmentation in complex environment,some effective improvement measures are put forward.For license plate location,in order to solve the problem that the license plate location method based on color statistics is easily affected by illumination conditions and the location accuracy is not high,an illumination compensation method is proposed.The vehicle image is converted from RGB color space to HSI color space,and the three components are separated.The illumination component contained in the luminance component is extracted by multi-scale Gaussian function.Then two-dimensional gamma function is constructed and adaptive brightness correction is carried out for the illumination component.Finally,it is converted back to RGB space,and the license plate location is carried out by the method based on color statistics.The experimental results show that for the common blue plate,this can not only well reduce the influence of light changes on the image,improve the quality of the image,but also can achieve the night and uneven light,complex background of the vehicle accurate positioning.To solve the problem that the license plate location method based on color statistics could not accurately locate the license plate when the color of the license plate is closer to the car body,a license plate location method based on color statistics and improved Canny operator is proposed.Firstly,the Canny operator is improved and a new formula for calculating the first partial derivative is proposed.Then the improved Canny operator is used to detect the edge of the license plate image and highlight the character region.Finally,after mathematical morphology processing,all candidate regions are obtained,and the location of license plate is obtained by analyzing them again.The experimental results show that the improved Canny operator can detect the vertical edge of the license plate well,reduce the transverse edge,and highlight the character texture.The proposed method can successfully locate the car body which is close to the color of the car body,and the locating accuracy is high and the locating speed is fast.For character segmentation,in order to solve the problem of horizontal slanting of license plate,a new method of horizontal slanting is proposed based on character enclost rectangle and Angle histogram.After processing the license plate image,multiple connected domains are obtained.After eliminating the interference areas,the character encasing rectangle is obtained.Then,the Angle of the character encasing rectangle is calculated and projected to the histogram to obtain the approximate inclination Angle.In order to solve the problem of character segmentation failure caused by character adhesion and disconnection of some Chinese characters,a character segmentation method based on the position feature of spacer and vertical projection is proposed.On the basis of vertical projection,the position and size of Chinese characters are backward deduced to the left with the spacer as the segmentation point,and the letters and numbers are segmented to the right with the help of character length characteristics.The experimental results show that the problems of disconnection and adhesion of some Chinese characters are well solved. |
PDF Full Text Request