吳家麟Wu, Ja-Ling臺灣大學:資訊網路與多媒體研究所王奕棠Wang, Yi-TangYi-TangWang2010-05-052018-07-052010-05-052018-07-052008U0001-2407200817070600http://ntur.lib.ntu.edu.tw//handle/246246/180618在這篇論文中，我們提出了一個系統，以在監控影片當中偵測心肺復甦術的施行並衡量其施行品質。此系統可以應用在急診室醫療團隊的培訓上，藉由系統的輔助，醫院中的指導醫生可以更有效地評估影片中之醫療團隊的心肺復甦術施行品質。我們先從影片當中抽取物體移動量的資訊，再利用這個資訊搭配支持向量機來判斷一個影片片段包含心肺復甦術與否。接下來我們會對包含心肺復甦術的片段做進一步的分析，並將分析所得的相關資訊嵌入影片。除此之外，為了避免侵犯病人的隱私權，我們會利用一個馬賽克的機制來遮蓋住病人身上露出的皮膚區域。我們拍攝了一些模擬影片來對這個系統進行測試，並且我們在心肺復甦術的偵測及施行品質的衡量上都獲得不錯的效果。In this thesis, we propose a system for automatically detecting cardiopulmonary resuscitation (CPR) events and analyzing CPR qualities from a surveillance video. The system is further applied to the training of healthcare providers in the emergency room. Instructors could more efficiently evaluate the CPR quality performed by a medical team with the aid of our system. We extract motion vectors from all image blocks and take motion information as a clue to classify video sequences into CPR and non-CPR segments based on the Support Vector Machine (SVM). We further analyze several indicators of CPR quality and attach CPR information to the video. In order not to infringe the privacy of the patient, we apply a mosaic mechanism to mask the skin color regions of the patient. Our system has been applied to several simulated CPR video sequences and the results show acceptable accuracy in CPR detection and CPR information measurement.1 INTRODUCTION 1 RELATED WORK 4.1 VIDEO SURVEILLANCE SYSTEMS 4.1.1 Intelligent Visual Surveillance Systems 4.1.2 Privacy Preserving 5.2 CARDIOPULMONARY RESUSCITATION 6 SYSTEM OVERVIEW 8 VIDEO ANALYSIS AND CPR DETECTION 10.1 MOTION SEQUENCE EXTRACTION 10.1.1 Motion Estimation 10.1.2 Group Motion Sequence 11.2 MOTION SEQUENCE CLASSIFICATION 13.2.1 Peak Detection 13.2.2 Feature Vector Generation 16.2.3 SVM Classification 17.3 HIERARCHICAL DETECTION 20.3.1 Group-level Classification 20.3.2 Frame-level Detection 22.3.3 Segment Boundary Refinement 23.4 CPR INFORMATION REPORTING 27.4.1 Compressions Delivered Per Minute and Compression Rate 27.4.2 Duration, Hand-off Time and Operator Change 29 PRIVACY PRESERVING 31.1 NON-MOVING SKIN COLOR 32.2 MOSAIC 33 EXPERIMENTAL RESULTS 35.1 CPR DETECTION 36.1.1 CPR Region 36.1.2 CPR Segment 37.2 CHEST COMPRESSIONS 38.3 PRIVACY PRESERVING 39 CONCLUSIONS AND FUTURE WORK 41EFERENCE 43application/pdf1847789 bytesapplication/pdfen-US心肺復甦術偵測心肺復甦術品質衡量支持向量機監控影片分析隱私權保護CPR detectionCPR quality measurementmotion sequence classificationsurveillance video analysisprivacy preservingthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/180618/1/ntu-97-R95944015-1.pdf