一個基於眼動的影片分析系統

A Video Analysis System based on Eye-Tracking

陳明道(Ming-Dao Chen)

指導教授 ： 袁賢銘

國立交通大學/資訊學院/網路工程研究所/碩士(2016年)

摘要

本研究提出了一個基於眼動的影片分析系統。為了讓影片製作者分析影片觀看者是否凝視影片中重要物件並判斷影片的優缺，同時感受到使用影片分析系統的便利性，我們改進影片分析系統 [6]，主要聚焦於兩大功能的演進:「動態興趣區模組」以及「影片分析」。
在我們演進的系統中，藉由眼球追蹤儀與「眼動紀錄」的結合，我們能記錄受測者的眼動資訊。「動態興趣區模組」提供使用者兩種設置模式－「手動設置模式」和「自動設置模式」。根據影片的特性，使用適合的動態興趣區設置模式可以更簡潔方便地定義影片中的動態興趣區。「影片分析」提供使用者三種模式－「單點」，「多點」，「熱圖」模式。根據分析的需求，提供單人或多人分析與不同的視覺化撥放並輸出簡明的統計圖表以及詳細的凝視資訊分析。
我們也實際讓受測者操作系統並填寫系統適用性問卷。根據問卷結果顯示我們的影片分析系統有著方便使用以及介面平易近人的特性。

關鍵字

眼動研究 ； 動態興趣區 ； 物件追蹤 ； 人臉辨識 ； 熱圖

並列摘要

This thesis proposes a video analysis system based on eye-tracking. In order to make video-maker to intuitively analyze that whether the video watcher fixates on important object in the video or not and to analyze the video, and to feel convenient when using video analysis system, we improve the system [6], mainly focus on two functionalities: “Dynamic AOI Module”, and “Video Analyzer”.
In our improved system, by linking eye tracking device with “Eye Movement Recorder”, we can record the subject’s eye movement data. “Dynamic AOI Module” provides two AOIs setting modes - “Manual Setting mode” and “Automatic Setting mode”. According to the video property, using appropriate AOI setting mode could conveniently define the dynamic AOIs in the video without more operation. “Video Analyzer” provides three modes - “Singular Spot mode”, “Plural Spot mode”, “Heat Map mode”. Depending on the requirement of analysis, it provides single or multiple subject analysis, different visualization displaying and output simple statistical charts and detail fixation information analysis.

並列關鍵字

Eye Tracking Technology ； Dynamic Area Of Interest ； Object Tracking ； Face Detection ； Heat Map

參考文獻

[1] D. Kumar and A. Sharmal, "Electrooculogram-based virtual reality game control using blink detection and gaze calibration" Communications and Informatics (ICACCI), International Conference on Advances in Computing, 2016

[2] S. I. Ktena, W. Abbott and A. A. Faisal, "A virtual reality platform for safe evaluation and training of natural gaze-based wheelchair driving" Neural Engineering (NER), 7th International IEEE/EMBS Conference on Neural Engineering, 2015

[3] P. Kasprowski, K. Harezlak, and M. Niezabitowski, "Eye movement tracking as a new promising modality for human computer interaction" Carpathian Control Conference (ICCC), 2016 17th International

[6] Z. Y. Peng, S. M. Yuan, "An Advertisement Video Analysis System based on Eye-Tracking," NCTU Library, 2014.

[7] H. Collewijn, F. Van der Mark, and T. Jansen, "Precise recording of human eye movements," Vision research, vol. 15, pp. 447-450, 1975.