機器視覺問卷輸入影像系統開發

Development of the Automatic Questionnaire Input System by Machine Vision

蒲永仁(Yong-Ren Pu) ； 李素幸(Su-Hsing Lee) ； 董思廷(Si-Ting Dong)

《勞工安全衛生研究季刊》 20卷1期 (2012/03) Pp. 70-77

https://doi.org/10.7005/JOSH.201203.0071

摘要

對許多職業安全衛生領域的研究者來說，透過問卷調查來進行定量訊息的收集是相當重要的，在大型的問卷調查中為了讓問卷分析的資料能有較高的準確性和代表性，數據的收集與管理往往會相當龐大，也使得問卷輸入成為研究過程裡的重要關卡。本研究發展一套資料收集的技術，可運用在職業安全衛生或其他學科的課題上，以提高問卷調查資料輸入的準確率及效率。本研究問卷辨識系統整合多項軟硬體模組，問卷經由送紙裝置逐一送出，透過攝像鏡頭擷取問卷的影像，利用圖控式程式語言，進行即時影像處理，並將分析後數據儲存成檔案。在準確率實驗中則針對筆的顏色及劃記形狀進行測試，經統計分析顯示黑、紅、藍、綠及鉛筆等大眾常用顏色之辨識率並無顯著差異；打勾及劃斜線之辨識率顯著高於劃叉或劃圈，整體而言，各顏色之辨識率均高達99.63％～99.77％，而各形狀之辨識率亦達95.7％～99.9％。此外分別對空白及填答問卷進行辨識，其辨識平均時間分別為627.82ms和785.5ms，即在0.8秒內軟體可以辨識一張填畢問卷。這些初步成果展現出快速及準確的辨識功能，為問卷輸入影像系統的開發奠定了良好的基礎。

關鍵字

問卷 ； 機器視覺 ； 影像處理

並列摘要

For many researchers in the area of occupational safety and health, it is very important to collect the quantitative information though questionnaire surveys. Hoping that the data of the questionnaire analysis can be more accurate and representative, people always collect and control a huge amount of data in a large-scale survey. Therefore, the questionnaire input becomes an important step in the analytical process. This study is designed to develop a technology to collect data for the occupational safety and health or other issues by improving the accuracy and efficiency in the questionnaire input. This developing system integrates a number of hardware and software modules. Each questionnaire sent by a paper feeder is captured by a camera, and analyzed in the graphical user interface in real time. The data acquired after image processing are then saved as a text file. In the first two experiments, we found by statistical analyses that the accuracies of the software to recognize the checkboxes filled with checks in various colors ranged from 99.63％ to 99.77％; and that to recognize other checkboxes answered with different marks were around 95.7％～99.9％. In the last experiment, we identified that the average processing time for a blank questionnaire was 627.82ms, and for an answered one was 785.5ms, which means that the developed software is capable of processing a questionnaire during 0.8 second. These preliminary results demonstrate that the software functions in a speedy and accurate manner, which has laid the firm groundwork for the automatic questionnaire input system in the near future.

並列關鍵字

Questionnaire ； Machine vision ； Image processing

參考文獻

Oppenheim, AN(1992).Questionnaire design, interviewing and attitude measurement.Pinter Publishers Ltd.

(Poor DDS. Image capture and storage techniques in association with optical mark reading. United States Patent 1995; 5,452,379).

Gonzalez, RC,Woods, RE(2002).Digital image processing.New Jersey:Prentice Hall.

(National Instruments Corporation (US). IMAQ vision concepts manual. Austin (TX): The Corporation; 2005.).

Xu, WR,Zhang, HG,Guo, J,Chen, G(2002).Discrimination between printed and handwritten characters for cheque OCR system.Machine Learning and Cybernetics.2,1048-53.