以階調修飾改善圖像化二維條碼之顯示效果

Abstract

隨著網路科技蓬勃發展與智慧型手機的普及，在日常生活中隨處可見各種媒體廣告雜誌及商品上由黑白方格所組成的QR code (Quick Response Code)，這種透過QR code來傳達資訊或連結網址的模式，逐漸成為虛擬網路與實體世界連結的重要媒介。ㄧ般QR code都是由黑白編碼組成，不甚美觀；有鑑於此，有許多學者相繼投入QR code與圖像整合研究，透過將圖像融於條碼來產生美觀並具有意義之QR code。現有的圖像化二維條碼，雖然可以明顯地辨別出當中的圖案，但是，仍然會有原加密二維條碼殘留的黑白點，影響視覺美感。本研究發展出調整灰階影像階調前處理程序，利用轉換函數(transfer function)修正影像亮部及暗部階調，設定一個截切臨界值C，當輸入影像灰階值低於C值或高於 255-C時，將動態範圍作裁減，為了能適當保留亮及暗部階調變化，以低頻通過濾波器將轉換函數做平滑處理，找出最佳灰階影像階調，再透過數位半色調技術之誤差擴散結合資訊隱藏演算法，在半色調進行過程中同時把二維條碼資訊藏入外顯影像，產生圖像化二維條碼。最後利用圖像化二維條碼的每一像素與其周圍的複數點的最近距離之分布圖及常態分配曲線圖之標準差σ值，來評估改變轉換函數影響QR code黑白點的遮掩效果，結果顯示二維條碼之有規律的黑白點，已被成功干擾，且所得遮掩效果較佳之圖像化二維條碼。As Internet technology and smart devices start becoming inseparable from our lives, we start to see an omnipresence of the QR code (Quick Response Code), a black-and-white dotted block which stores information and allows quick access to Internet URLs. Since regular QR codes are not aesthetically pleasing, many experts began to look for ways to integrate QR codes with images to generate graphic QR codes. However, these graphic QR codes are still left with traces of patterns that could undermine its aesthetic values. To solve this problem, this paper aims to explore the effectiveness of a pre-processing program to adjust the gray-scale image tone, using a transfer function to correct the image tone of highlights and shadows. After setting a parameter value C, the dynamic range will clip the image if the gray-scale value is either a) lower than C or b) above 255-C. To maintain a proper tone scale from highlight to shadow, a low pass filter is employed for image smoothing to gain the optimal gray-scale image tone. Then, by combining the error diffusion function and data-hiding algo-rithm, QR code information can be embedded into images during halftoning process to generate a graphic QR code. Finally, we can evaluate the effectiveness of QR code pattern obscuration by referencing the σ value, which is the standard deviation be-tween the proximity distribution and normal distribution curve of each pixel and its surrounding dots. The results show that the regular distribution of the pattern has been successfully interfered, and therefore we obtain a graphic QR code with a better hiding effect.