陳世旺梁祐銘Sei-Wang ChenYu-Ming Liang陳俊宇2019-09-052014-10-12019-09-052014http://etds.lib.ntnu.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dstdcdr&s=id=%22GN060147040S%22.&%22.id.&http://rportal.lib.ntnu.edu.tw:80/handle/20.500.12235/106606隨著科技的進步及攝影器材的普及，視訊監控已成為我們生活中最重要的安全監控工具之一，而車種分類與計數更在智慧型交通安全監控系統中扮演著重要的角色，其目的是希望能改善交通壅塞與安全的問題。本研究發展一個以電腦視覺為基礎的即時車種分類及車輛計數系統。在系統運作的過程中，主要可以分為兩個大步驟，第一大步驟為車輛的擷取，第二大步驟為車輛分類與計數。在車輛的擷取的部份，首先對輸入的影片建立Time-Spatial Images(TSI)，並利用Support Vector Machine(SVM)與HSV Based on Deterministic Non-Model Based Approach來分類陰影與非陰影，將TSI圖中的陰影部份去除後，再透過簡單的morphology處理擷取Region of Interest(ROI)即為車輛在TSI圖的區域。在車輛分類與計數的部份，我們使用ROI累計曲線法和Fuzzy Constraints Satisfaction Propagation(FCSP)演算法來處理遮蔽的問題，利用Fuzzy的觀念進行各種車種模糊比對，從有遮掩情形的ROI區域中分離出獨立的車輛，並進行車輛的分類與計數。實驗的結果顯示所提技術可以在無特殊輔助硬體的環境下，能有效且即時地執行車種分類與計數，並證明了此方法具可行性的。Vehicle classification and counting play an important role in the intelligent transportation system, as they may serve to improve traffic congestion and safety problems. Therefore, this study has developed a real-time and computer-based visual vehicle classification and counting system. This will involve establishing Time-Spatial Images (TSI) from input video, removing the shadow portions in TSI through the use of Support Vector Machine (SVM) and Deterministic Non-Model Based Approach, capturing the Region of Interest (ROI) through a simple morphology process, and finally using the ROI accumulative curve method and Fuzzy Constraints Satisfaction Propagation (FCSP) to process occlusion problems and perform vehicle classification and counting. The experimental results have shown that the proposed method is feasible.累計曲線法Time-Spatial ImagSVM陰影去除模糊限制滿足技術遮蔽處理Accumulated CurveTime-Spatial ImagSVM Shadow RemovalFuzzy Constraints Satisfaction Propagationocclusion