在本論文中,我們使用雙通道Mach-Zehnder聯合轉換相關器在YUV色彩模型下對多彩圖像的辨識。多通道聯合轉換相關器被提出在三通道上辨識多彩圖像。然而,它需要較大的液晶空間光調制器來對多彩圖像辨識。YUV色彩模型的特性有利於只使用兩個多彩色差元素來辨識。此外,我們使用最小平均相關能量的方法來完成形變不變性。多階量化參考函數的方法被使用在濾波器的設計,使之更適合實現此光學相關器的輸入平面。接著,我們評估對於旋轉形變,量化程度,通道間的相關性,雜訊環境,包含大小與角度之形變,以及多目標圖像下的效果。最後,由數值結果得知,多彩圖像在YUV色彩模型的雙通道下可以成功的被辨識出來。
In this thesis, we utilize two-channel Mach-Zehnder joint transform correlator based on YUV color model for chromatic pattern recognition. The multi-channel joint transform correlator was proposed to discriminate the chromatic pattern on three channels. However, it requires a larger liquid crystal spatial light modulator for chromatic pattern recognition. The property of YUV color model contributes to utilize only two chromatic aberration components on recognition. Then, we utilize the minimum average correlation energy to achieve the distortion invariant. In order to be more suitable for realizing the input plane of optical correlator, the multi-level quantized reference function is applied to design the filter. Subsequently, we estimate the effects for rotated distortion, quantized level, correlation between channels, noisy environment, distortion with scale and angle, and multiple targets pattern. Finally, the numerical results of simulation show that chromatic pattern based on two channels of YUV color model is successfully detected.