傳統條碼辨識在光照不均的環境下，會造成條碼影像分割錯誤及辨識率不足的問題。本研究應用迭代式Otsu方法來改善影像的二值化，利用迭代搜索的分割影像的子區域，對於所採集的條碼影像，計算不同光照區域的最佳閾值，以增進影像分割的可靠度，從而提升條碼辨識率。我們實作了此方法，並以一組取樣品質不佳的條碼影像資料庫進行測試，測試結果表明，該方法確實對不同光照環境的影像能夠獲得更好的二值化結果。此一結果將能被應用到真實工業生產條碼的辨識。;Traditionally, barcode recognition rate will be affected by uneven illumination since it may cause the faults of image segmentation and then raise the error rate of recognition. In this paper, we try to use an iterative triclass thresholding technique proposed by Cai to improve the thresholding. The method uses iterative search for image segmentation and computes the optimal threshold values for different illumination regions. Thus, the reliability of image segmentation is increased and the rate of barcode recognition is improved. We implemented this method and tested it with a group of inferior sampling quality barcode images. The experiment data shows that the method actually obtains a better result of thresholding on images under the uneven illumination. The result will be applied to actually industrial barcode recognition.
