問卷輸入系統自動調整最佳攝距參數研究

Study of the Automatic Adjustment on the Optimal Shooting Range for Questionnaire Input System

蒲永仁(Yong-Ren Pu)；戴辰倫(Chen-Lun Tai)；李素幸(Su-Hsing Lee)

問卷辨識 ； 機器視覺 ； 安全衛生 ； Questionnaire Recognition ； Machine Vision ； Safety and Health

勞動及職業安全衛生研究季刊

26卷1期（2018 / 03 / 15）

28 - 37

繁體中文

問卷調查不但是現今各類科學研究蒐集資料經常使用的方法，在職業安全衛生領域中亦扮演著相當重要的角色，研究中為求分析的數據能有足夠的準確性及代表性，往往需要發放與回收相當數量的問卷，若利用機器視覺進行大量問卷的快速辨識，必會遭遇攝距逐漸降低的情形，以往需藉著手動調整攝距參數，才得以達到準確定位的目的。因此本研究之動機係在發展最佳攝距參數自動調整功能，利用迴歸分析找出定位標距離與最佳攝距參數之關係，接著進行程式設計，將其整合入原系統，使其具有精確定位的能力。辨識實驗結果之WSR test分析顯示，自動攝距調整功能，對於降低ROI框心與答案框心間之距離誤差像素值，具有統計上顯著的意義，因此能增加問卷辨識之準確性，使此系統能夠發揮更大的功效。

Questionnaire Surveys are often used to collect information in a variety of scientific research fields, especially in one of occupational safety and health. For the accuracy and representative of the collected information, people usually administer a considerable amount of questionnaires. Using machine vision to process a large amount of questionnaires, researchers would inevitably face the fact that the shooting distance will be decreasing as the process goes along. They needed to manually adjust the parameter of the shooting range in order to be able to locate the checkboxes therein. The goal of this work is to develop the automatic adjusting function of the optimal shooting range for the existing system. By generating a higher power regression equation between the distances of the locating marks and the optimal shooting range parameters, the system is programed with the automatic function in accordance with the equation. The WSR test on the experiment results show statistically that this developed automatic functions can effectively reduce the errors between the aiming ROI centers and the aimed checkbox centers. Therefore, the revised system not only can increase the accuracy of the questionnaire recognition, but also raise the efficacy for the software usage.

1. 蒲永仁、李素幸、董思廷(2012)。機器視覺問卷輸入影像系統開發。勞工安全衛生研究季刊，20，70-7。
   連結：
2. Huang, W,Tan, CL,Sung, SY,Xu, Y(2001).Vertical bar detection for gauging text similarity of document images.Document Analysis and Recognition
3. Milyaev, S,Barinova, O,Novikova, T,Kohli, P,Lempitsky, V(2013).Image binarization for endto-end text understanding in natural images.Proceeding of 12th International Conference on Document Analysis and Recognition,Washington, DC.:
4. Oppenheim, AN(1992).Questionnaire design, interviewing and attitude measurement.Continuum.
5. Pu, YR,Chang, YC,Lee, SH(2009).Development of a questionnaire input software by machine vision.Proceedings of 2nd International Symposium on Knowledge Acquisition and Modeling,Wuhan, China:
6. Pu, YR,Lee, SH,Kuo, CL(2011).Calibration of the camera used in a questionnaire input system by computer vision.Information Technology Journal,10,1717-24.
7. Pu, YR,Lee, SH,Wang, MH,Luo, YC,Kuo, CL(2014).Nonlinearity of radial distortion in a questionnaire input software by computer vision.Proceeding of 5th International Conference on Nonlinear Science and Complexity,Xi'an, China:
8. Riedl, C,Zanibbi, R,Hearst, MA,Zhu, S,Menietti, M,Crusan, J(2016).Detecting figures and part labels in patents: competitionbased development of graphics recognition algorithms.Internation Journal on Document Analysis and Recognition,19,155-72.
9. Rosner, B(2011).Fundamentals of biostatistics.Boston:Cengage Learning.
10. Tunney, WP(2006).Tunney WP. Method and system for identifying multiple questionnaire pages. United States Patent 7031520 B2; Apr. 18, 2006..
11. 林瑞明(2002)。碩士論文(碩士論文)。國立台中師範學院教育測驗統計研究所。
12. 曾文憲(2005)。碩士論文(碩士論文)。國立陽明大學衛生資訊與決策研究所。
13. 戴辰倫、蒲永仁、李素幸(2015)。問卷輸入系統最佳攝距參數迴歸分析研究。第五屆職業安全研討會，台灣、苗栗: