可調性視訊之人臉偵測

Face Detection in Scalable Video

林美亨(Lin Mei-Heng)

指導教授 ： 傅心家

國立交通大學/資訊學院/資訊學院碩士在職專班資訊組/碩士(2006年)

https://doi.org/10.6842/NCTU.2006.00001

摘要

本論文研製之 3D小波可調性編碼( 3D Wavelet Scalable Video Coding)擁有頻寬可調性、時間軸可調性與空間軸可調性等優點，可分為三個主要的模組:時間小波 T_Module、空間小波S_Module與Entropy_Coding模組以產生最後壓縮資料串流。本論文我們希望結合小波可調性編碼與人臉偵測系統，一次性嵌入人臉偵測於不同等級資料串中，針對不同的網路需求，讓接收端用最小資料量分析最快偵測出人臉影像。對可調性視訊接收端而言：資訊接收端不再受限於單一撥放環境，接收端可根據自身的網路環境，選擇觀看人臉偵測影像。對可調性視訊服務端而言：網路伺服器只需提供一個檔案供所有使用端下載，無論接收端的網路環境為何，對伺服器都只要求同一個檔案，不但可以提供不同使用端的需求，更可節省檔案伺服器對硬體需求上的負擔。

由實驗結果驗証，在不同時間壓縮、空間壓縮下之不同等級資料串流偵測人臉影像，均可達到可調性視訊人臉偵測目標，並達到最小資料量分析最快偵測結果。

關鍵字

可調性視訊 ； 人臉偵測

並列摘要

3D wavelet scalable video coding has the advantages of bandwith, temporal, and spatial scalability, which includes three main modules : T_Module、S_Module、Entropy_Coding, those produce final compressed output bit stream.

In this thesis we would like to combine face detection in 3D wavelet scalable video. The advantages are: 1. User only download one bit stream and extract it for what they need. 2. Server only provides one bit stream for all users to download and it also saves lots of disk space in network.

We propose a strategy that makes the face can be detected under different environment and detect the face correctly and quickly.

並列關鍵字

Scalable Video ； Face Detection

參考文獻

3.J.-Z. Xu, S. Li and Y.-Q. Zhang, “Three-dimensional shape-adaptive discrete wavelet transforms for efficient object-based video coding,” to be appeared in Proceedings of SPIE VCIP’2000, Perth, Australia, July, 2000.

4.Lin Luo, Feng Wu, Shipeng Li, Zhenquan Zhuang, and Ya-Qin Zhang, A two-pass optimal motion-threading technique for 3D wavelet video coding, IEEE Intl. Symposium on Circuits and Systems (ISCAS 2002), pp.819-822, Scottsdale, AZ

5.Lin Luo, Feng Wu, Shipeng Li, and Zhenquan Zhuang, Advanced lifting-based Motion-Threading (MTh) techniques for 3D wavelet video coding, invited paper, to appear in SPIE/IEEE Visual Communications and Image Processing (VCIP2003), Lugano, Switzerland

6.Lin Luo, Feng Wu, Shipeng Li, and Zhenquan Zhuang, Layer-correlated motion estimation and motion vector coding for the 3D-wavelet video coding, IEEE International Conference on Image Processing (ICIP2003), Barcelona, Spain

8.Wanzeng Kong , Shan`an Zhu ”,A New Method of Single Color Face Detection Based on Skin Model and Gaussian Distribution 2006”