1/1000數值地形圖圖檔資料內部圖元檢核之研究－－以苗栗縣公館與銅鑼鄉圖幅為例

A Study towards Graphic Elements Checking of 1/1000 Digital Topographic Maps – A Case Study of Gongguan and Tunglo Township

本研究從品質檢核的客觀性角度切入探討，並以苗栗縣1/1000 都市計畫航測數值地形圖立製編輯初稿之公館2904518 圖號及銅鑼2854515 圖號作為本研究之實驗測試區，其方式為先以傳統人工抽樣驗收之方式進行測試，將不符合「基本地形圖資料庫圖式規格表」的項目進行整理分類，從而做為開發程式檢核條件約制之方法依據。再與其它已經人工檢核驗收完成後之3 個測試區各單一圖幅，採用開發完成後之程式檢核進行測試比較。研究結果顯示實驗測試區採用人工判視檢核需投入4 小時／2 人才找到214 個錯誤量，而採用程式檢核僅需10 分鐘／1 人就可檢核出錯誤量達19852 個，並自動產生報表及展繪在圖檔上供檢核人員查看及統計，而其他3 個測試區經程式檢核其圖元錯誤量占總量之比例分別為26.8%， 76.2%及5.19%，本文研究成果可提供1/1000 數值地形圖圖元檢核之應用。 From objective view, this study selected drafts (No.2904518 and No.2854515) of 1:1000urban planning aerial photographic maps of Gongguan and Tunglo Township of Miaoli County asthe testing areas, and adopted conventional manual sampling for testing. Maps which failed tomeet the Format Form of Basic Topographic Map Database were classified and served as the basisof constraint conditions of inspection program. The 1:1000 maps (in testing areas A, B and C)inspected manually were compared with the results of inspection program.The obtained results from this study indicated that manual inspection in experimental arearequires 4 hours/2 persons to find 214 errors, and the inspection program only spends 10 min/1person to find 19852 errors, and generate report statistical analysis. The pixel errors of maps (in testing areas A, B, C) accounted for 26.8%, 76.2% and 5.19% of total errors respectively. Theobtained results from this study can be used in practical pixel checking of 1/1000 digitaltopographic maps.

From objective view, this study selected drafts (No.2904518 and No.2854515) of 1:1000urban planning aerial photographic maps of Gongguan and Tunglo Township of Miaoli County asthe testing areas, and adopted conventional manual sampling for testing. Maps which failed tomeet the Format Form of Basic Topographic Map Database were classified and served as the basisof constraint conditions of inspection program. The 1:1000 maps (in testing areas A, B and C)inspected manually were compared with the results of inspection program.The obtained results from this study indicated that manual inspection in experimental arearequires 4 hours/2 persons to find 214 errors, and the inspection program only spends 10 min/1person to find 19852 errors, and generate report statistical analysis. The pixel errors of maps (in testing areas A, B, C) accounted for 26.8%, 76.2% and 5.19% of total errors respectively. Theobtained results from this study can be used in practical pixel checking of 1/1000 digitaltopographic maps.