本研究中,最主要的概念是利用模擬退火法創出參考影像,並配合多通道去零階聯合轉換相關器在HSV色彩空間中對目標影像進行辨識。 我們將原始目標影像從 到 每 旋轉一次,並且儲存這些圖像當作訓練影像。接下來,便可以利用模擬退火法產生最佳化的參考影像。此外,機率參數必須適當地設定以產生全域最佳化的參考影像。最後,我們就可以使用這些參考影像配合去零階聯合轉換液晶相關器進行彩色影像辨識。
The main idea of this research is to find the optimal filter to recognize the target image, which has different distortions, by utilizing simulated annealing (SA) algorithm with multi-channel joint transform correlator on HSV color space. We rotate the original colorful images and save them from to in steps of as our training image set. Next, the optimal filter could be yielded by SA method. Additionally, in order to yield global convergence filter, the probability parameter should be set appropriately. Finally, we are able to use these filters to recognize the colorful pattern on nonzero order joint transform correlator (NOJTC).