以紅外線浮水印技術於圖像化二維條碼中隱藏訊息之研究

Abstract

隨著科技發展愈趨進步，各種資訊傳播交流更為快速，但仍須有一媒介連結實體與虛擬的世界，而二維條碼因其可被快速解碼而廣為使用，其中最具代表性的二維條碼為QR Code (Quick Response Code)。可是目前市面上的QR Code僅供機器判讀，於人眼視覺觀看並不美觀，許多學者亦投入美化QR Code之相關研究。另外由於透過QR Code可連結至相關網站或作為交易支付之用途，若遭有心人士惡意盜用或竊取個人資料將會造成使用者的損失，可見QR Code的安全機制仍有強化之必要性，因此本研究將使用紅外線浮水印技術於圖像化二維條碼中藏入第二個QR Code，以提高圖像化二維條碼之安全性。本研究首先以資訊隱藏技術結合誤差擴散法，於圖像中藏入原始QR Code之資訊，形成圖像化QR Code。另外為了避免外顯圖像化QR Code在輸出過程中產生黑墨影響到紅外線浮水印，會將外顯圖像化QR Code之黑墨成分資訊分配至C (Cyan)、M (Magenta)、Y (Yellow)三色版中，而欲隱藏之第二個QR Code資訊則以紅外線浮水印技術置於CMYK模式中的K (Black)層，並加入了30到120等不同黑墨灰階值之浮水印，使其可在不影響外顯圖像化QR Code判讀情況下，以紅外光照射仍可取得浮水印資訊，最後將兩者結合後輸出即可形成一個具有防偽功能之圖像化二維條碼。若非法複印此圖像化QR Code，則複製品於紅外光照射下將無法顯現隱藏之QR Code。另外再以資料點辨識程式計算圖像化QR Code之辨識錯誤率。結果顯示，計算資料點的辨識錯誤率介於0.6%至12%之間。而隱藏的紅外線浮水印，當黑墨灰階值強度為50至120之間即可以手機直接判讀紅外線儀器偵測到之QR Code資訊，並可在不影響圖像化QR Code美觀度及解碼性的前提下，提升二維條碼的安全性。

With the development of technology, the information delivers much faster than before. However, a link to connect the virtual and physical world is still mandatory. And the most representative one is the Quick Response Code (QR Code). However, in the condition for the current market, QR Code can only be decoded by machine, which makes it not so beautiful for human's sense of sight. Therefore, many researchers have been involving in the research of graphic QR Code. Nowadays there are many stores offer QR Code for transactional payments. If someone tries to duplicate other's QR Codes, then the transaction safety of the consumer might be concerned. As a result, the security of the QR Code still needs to be strengthened. Therefore, the study used infrared watermark technology to hide the second QR Code in the graphic QR Code, with the purpose of improving the graphics QR Code security. The study first produced graphic QR Codes, the second QR Code embedded in the graphic QR Code by infrared watermark technology with the combination of a serial number of different intensities. Finally, the recognition error rate of QR Code data points is calculated. As for the production of explicit graphic QR Code, it hides the data in the CMY layer to prevent the black ink (K) from affecting the infrared watermark. The second QR Code and its serial number are placed in the K layer of the QR Code of the image. Then the combination of the two output can be formed with a security feature of the graphic QR Code. If someone copies this image QR Code illegally, the replica will not be able to show the hidden QR Code under the infrared light. So as to understand the influence of different infrared watermarking intensities (30 to 120) on the decoding of explicit graphic QR Code, we will print the infrared watermarking graphic QR Code and scan it to calculate the recognition error rate of the graphic QR Code. The results show that the error rate of explicit graphic QR Code data points is between 0.6% and 1.2%, while the infrared watermark QR Code data can be successfully decoded at intensities of 50 to 120. Without compromising the appearance and decoding of the QR Code, the security of the QR Code is improved.