| Summary: | 碩士 === 國立高雄第一科技大學 === 電腦與通訊工程系碩士班 === 105 === Barcode has been widely used in the logistics industry. The EAN-13 barcode has very large demand, so it is important for us to obtain the product information through the automatic recognition of EAN-13 barcode. The image recognition method can save the cost when it is compared with other identification methods. And, the image captured by the camera is also flexible in various environments. Thus, the purpose of this thesis is to study the bar code image decoding and the bar code digit number recognition.
In this thesis, the features of bar and digit-number are separately extracted such that EAN-13 code recognitions of bar and digit-number can be individually developed. In final barcode recognition results, bar decoding result is the major choice, the result of digit-number is the minor choice. Proposed method uses two object location methods, one tilt correction, five pre-processing methods on the bar, three pre-processing methods on digit number, and three binarization methods to segment the foreground and background. The traditional barcode construction rule is employed to decode the bar. Optical character recognition (OCR) methods have been developed for many years, so the k-nearest neighbors (KNN) algorithm, template similarity matching, and support vector machine (SVM) are used to recognize the digit-numbers.
The combination of bar and digit-number recognition results can overcome many difficulties in barcode automatic identifications, so the proposed method is a promising approach to recognize EAN-13 barcode. Experimental results show that the proposed method achieves the recognition rate of 85% and the average recognition time is 1.25 seconds.
|
|---|
