本論文中我們提出了一個新的方法,應用在聯合轉換相關器的彩色影像識別系統。此技術是利用影像編碼將每一個彩色目標影像轉換到灰階影像。當輸入為灰階或單色時,就可利用液晶光調制器來達到及時性的運作。在另一方面,我們使用最小平均相關能量的方法,去設計一個最佳化的合成參考函數。除此之外,我們也利用功率頻譜相減的方法,來移除輸出面上的零階項,使預期的相關峰值看得較明顯。 為了證實我們所提出的方法是有成效的。最後,我們藉由電腦的數值計算來測試評估雜訊及形變的目標影像,如PCE、PSR、及CPIR等。此時我們發現新的聯合轉換相關器擁有較高的影像辨識力和偵測能力。
In this thesis, we propose a novel method for color pattern recognition based on the joint transform correlator architecture. In the method each color target image is transformed to a grayscale image with image encoding technique. The input plane is gray-scaled and monochromatic, thus it can be displayed in the liquid-crystal spatial light modulator to achieve real-time operation. On the other hand, we use minimum average correlation energy approach to design an optimized synthetic reference function. Moreover, we adopt the joint transform power spectrum subtraction strategy to remove zero order part at the output plane, and the desired correlation peaks can be easily detected. At the end, we study system’s performance for noise and distorted images numerically. Thus, we show that the novel joint transform correlator has higher pattern discrimination ability and detection ability.