車牌號碼、條碼、貨櫃碼自動辨識方法

• Main Authors: Cheng-Yu Wen, 温震宇
• Other Authors: Chien-Cheng Tseng
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/46041673924829990639

碩士 === 國立高雄第一科技大學 === 電腦與通訊工程研究所 === 104 === In recent years, optical character recognition (OCR) has been widely used in intelligent transportation system (ITS). Traffic and flow of goods can be effectively managed by using the automatic recognitions of license plate number, barcode, and container code. Automatic recognition method is low-cost and safe, so this thesis studies how to stably recognize the alphanumeric in various environments. In this thesis, six color image enhancement methods are first compared. The high-speed color saturation method is selected to enhance the night-time license plate image and improve the contrast of barcode image. After performing enhancement and tilt correction, we can get better performance of localization and segmentation of night-time license plate images. And, the support vector machine (SVM), artificial neural networks (ANN) and Euclidean distance are combined to recognize the alphanumeric. In the barcode recognition, we first utilize the gradient to extract the barcode region. Then, we sharpen and enhance the contrast of barcode region to solve the blur problem caused by the tremor and minimum focal length. Next, the Hough transform is used to find the horizontal axis and correct the tilt angle. Finally, we find out the location of numbers such that barcode can be segmented and recognized. In the container code recognition, we directly use the Bernsen binarization method to separate foreground from background. Then, we use size filter, support vector machine and median filter to remove noises. Finally, the Otsu global method is used to binarize the segmented characters and perform character recognition. The proposed image enhancement method can provide better visual quality. And, the recognition rates of three automatic recognition methods are high. Thus, the proposed recognition methods can be effectively used in the vehicle parking, traffic transportation and flow of goods.