應用腦波偵測於演化策略建立嫌犯辨識系統

Incorporated event-related potentials brain waves and evolution strategies to establish criminal recognition system

林姿儀(Tzu-Yi Lin)

指導教授 ： 張應華

淡江大學/商管學院/資訊管理學系碩士班/碩士(2017年)

https://doi.org/10.6846/TKU.2017.00706

摘要

近年來，在人臉辨識方面利用上已相當廣泛，辨別人是否處於疲累狀態、出入口管制及利用辨識技術，辨識出人與物體，可以進一步辨別是否有人犯罪等。另外，腦波的應用發展也相當迅速，腦波現階段可以用來分辨嫌犯是否說謊、學生對學習是否專注、在醫學上判斷是否有腦部疾病以及在運動上可幫助運動員找到最適合比賽時的狀態等。
現今用於嫌犯之人臉辨識大多處於須連線比對資料庫的情況，受害者或目擊者欲描述其印象中的嫌犯特徵時，需與警方素描師反覆確認是否為嫌犯，此過程反覆時間長，並且有時受害者或目擊者所描述的與嫌犯特徵有大部分不同，進而影響警方作業速度，或是警方在街頭遇到嫌犯或失蹤人口時，欲比對是否為嫌犯或失蹤人口，必須連線到資料庫進行比對，此過程還須等待資料傳輸時間，當網路不順或不通時，將無法即時確認是否為嫌犯或失蹤人口。
因此，本研究欲利用演化策略整合事件相關電位中的腦波以發展出一套嫌犯描繪系統，利用腦波事件相關電位的擷取系統，目擊者及受害者不需提供犯罪者的特徵，經由取得使用者之腦波事件相關電位，在系統中解出嫌犯之特徵由演化策略快速找出犯罪人素描圖，將結果顯示於系統畫面，由使用者確認是否與印象中之長相符合。
此系統可以增加警方作業速度，縮短原本素描師與目擊者及受害者反覆確認所耗費之時間，也可利用於一般使用者，欲要告知他人印象中之人物的特徵，可透過此系統，利用偵測腦波事件相關電位，得出想要描述之特徵或人物。

關鍵字

事件相關電位 ； 演化策略 ； 類神經演算法

並列摘要

The face recognition are used quite extensive in recent years, like identify whether people are in a state for exhaustion, import ane export control and used to distinguish people or objects, can also identify whether someone in the process of committing a crime. In addition, the application of brain waves is development quite rapidly. Brain waves can be used to distinguish whether the suspects are lying and the students focus on learning. In terms of medical can judge whether has brain disease. In terms of sports can help athletes find the most suitable state of competition.
At present, the most of face recognition of suspect are connect database to compare. The victims or witnesses were describe the features of suspects confirmed repeatedly with police sketcher. This process confirmed repeatedly and time-consuming. If the victims or witnesses described the facial features incomplete resulted in suspect's sketches were different from suspect and then affect the speed of police operations. Or the police encountered the suspect or missing persons want to be compared whether the person was the suspect or missing persons, it must be connect database to compare pictures of the suspect or missing persons, the process must wait for the data to be transmitted. When the network has bad connection, it will not to confirm whether it is a suspect or missing persons immediately.
Therefore, this project intends to use the evolutionary strategy and neural network integrate with event-related brain waves in order to develop the identification system of suspects of crime. The use of event-related potential brain waves captured, witnesses or victims were no longer to describe the features of suspects. In the system, by obtaining the brain wave event-related potential of the user, evolutionary strategy and neural networks can be used to find the criminal sketch quickly without the description of the characteristics for the offender. The result can be displayed on the system screen. And confirm the result correct or not by user.
This system can increase the speed of police operations, shorten the sketch and witnesses or victims confirmed repeatedly and time-consuming, can also be used for general users. Through this system, by detecting the brain waves event-related potential and use evolutionary strategy to describe who you want to tell others the features of the character in the impression.

並列關鍵字

Event-Related Potential Brain Waves ； Evolutionary Strategy ； Artificial Neural Networks

參考文獻

[6] 吳明芳、李振興、王炳聰、詹慧珊、黃建邦，〈多人臉影像視覺辨識技術〉，崑山科技大學學報，第8期，頁75-95，2011。

[11] 林豐澤，〈演化式計算上篇：演化式演算法的三種理論模式〉，智慧科技與應用統計學報，2005。

[15] 高士竣、洪聰敏、黃崇儒，〈較佳精準運動表現中專注的腦波特徵〉，中華體育季刊，第23卷，第3期，頁1-16，2009。

[18] 張佐文，〈影像討論－腦波簡介〉，慈濟醫學雜誌，第1卷，第3期，頁85-89，1989。

[21] 張財榮、賴建庭，〈混合類神經網路應用於影像車號辨識〉，資訊科技國際期刊，第10卷，第1期，頁100-111，2007。