| 摘要: | 本論文中，我們提出以二維對稱遮罩式離散小波轉換及人臉狀態查表法應用於高效能人臉偵測與遮蔽處理系統。利用離散小波轉換後其低解析度影像的小波臉特徵擷取以及搭配特徵狀態查表法以增加其系統之運算複雜度。由實驗結果得知，本文所提出的方法可以在室內環境時的多人臉與不同角度下均達到即時處理並且保持較高的人臉偵測率。
In this work, a highly efficient face detection and face occlusion system with Symmetric Mask-based Discrete Wavelet Transform (SMDWT) and facial condition Look-up-Table (LUT) is proposed. It uses a low resolution image and the LUT to improve the face detection system to achieve real-time processing, and also adopts the wavelet face feature detection method to increase correction ratio. Our approach has been verified on captured images under various real indoor environments. As a result, it can provide a higher correction rate in the detection of multiple faces for real-time face detection applications. |
