陳世銘Chen, Suming臺灣大學:生物產業機電工程學研究所楊宜璋Yang, I-ChangI-ChangYang2010-05-052018-07-102010-05-052018-07-102009U0001-0907200918560600http://ntur.lib.ntu.edu.tw//handle/246246/180248使用非破壞、非侵入性的檢測方法已漸成為生物材料測定方法的新趨勢。而光學方法正符合這樣的性質，且能依據不同的檢測目的來進行檢測系統的設計，包含外部與內部品質。本論文包含三部份，從光學分析的角度，探討穀物的安全及品質檢測，並延伸探討在溫室中種苗生長的檢測與管理。在第一部份提到鐮胞菌枯萎症 (Fusarium Head Blight, FHB) 是一種在小粒穀物上由真菌引起之全球性疾病，因此影響了產量、食品和飼料產品的品質與食用安全。本研究發展有效的光學檢測方法，從正常的麥粒中，檢測枯萎損壞的小麥。經由高功率脈衝的LED （綠光和紅光）檢測系統的發展，發現鐮胞菌損壞和正常小麥粒有不同的反射能量表現，以其反射響應強度迴歸而得之雙參數（斜率與判定係數r2）建立線性判別分析模型。在良好、受控制的條件下，麥粒以動態自由落體方式檢測其精度，可大於90％。第二部份研究中，應用高光譜影像來鑑別健康小麥與由真菌引起的黑頭損傷。經由波長影像之分析，於531 nm 波長取得之螢光影像用於影像處理及分類之分析。分析結果指出，當麥粒皆位於背向朝著上方攝影機時，其分類的準確率可高達95%。論文的第三部份，是將RFID（無線射頻辨識系統）導入溫室內精準栽培環境，以光譜影像遙測技術為基礎，針對溫室開發多功能監測系統。透過RFID技術建立一套種苗生產履歷資料庫，內部架構涵蓋栽培管理履歷、設施環境履歷。對溫室內種苗植株進行遙測光譜影像及環境因子擷取，搭配RFID技術整合生產相關履歷資料，自動巡走及擷取、記錄相關種苗生產資訊，以提供溫室種苗生產管理之依據。Non-destructive and non-invasive inspection of biomaterials is a relatively new technology. The detection apparatus design depends on the purpose of the inspection, such as whether external or internal quality is of interest. There are three parts in this dissertation. From the view point of the optical analysis, the safety and quality issues of grain were discussed in the first two parts, and the monitoring and management of seedling production in the greenhouse was also elaborated in this dissertation. These three parts are summarized as follows: First, a study was implemented to develop more efficient methods for optically detecting wheat kernels damaged by Fusarium head blight, a fungal disease that is usually accompanied by the mycotoxin, deoxynivalenol. Through development of a high-power pulsed LED (green and red light) inspection system, it was found that Fusarium-damaged and normal wheat kernels have different reflected energy responses. Two parameters (slope and coefficient of determination r2) from a regression analysis of the green and red LED responses were used as input parameters in linear discriminant analysis models. The examined factors affecting accuracy were the orientation of the optical probe, the color contrast between normal and Fusarium-damaged kernels, and the manner in which one LED’s response was time-matched to the other LED. The current research on free-falling kernels has demonstrated accuracies (>90% for wheat samples with high visual contrast) that approach those of controlled, in-laboratory conditions. This approach may lead to improvements in high-speed optical sorters. Second, a feasibility study was conducted on the use of hyperspectral imaging to differentiate sound wheat kernels from those with a damage condition called black point. Through analysis of wavelength images, one fluorescence wavelength (531 nm) was selected for image processing and classification analysis. Results indicated that with this wavelength alone, classification accuracy could be as high as 95% when kernels were oriented with their dorsal side toward the camera. Third, a system was designed and implemented for precision cultivation and micro-environment monitoring for seedling production in the greenhouse. Based on RFID-integrated multi-functional remote sensors with a plant-oriented sensing algorithm for both monitoring and controlling the greenhouse environment, the system linked spectral and environmental inputs for the control of seedling irrigation. Further, the study aimed to construct a traceability system for seedling production in greenhouses using RFID technology. The contents of the developed database were divided into two parts, a management traceability system and an environment traceability system. The traceability systems provided the operators with records of seedling growth and management history and served as the decision bases for spray and related operations.口試委員會審定書 i謝 iiCKNOWLEDGEMENTS iii文摘要 ivBSTRACT vONTENTS viiIST OF FIGURES xiiIST OF TABLES xviiIST OF SYMBOL xviiiHAPTER 1. GENEREAL INTRODUCTION 1.1 INTRODUCTION 1.2 PROJECT BACKGROUND 7.3 GENERAL OBJECTIVE 7.4 DISSERTATION ORGANIZATION 7HAPTER 2. ENHANCEMENT OF FUSARIUM HEAD BLIGHT DETECTION IN FREE-FALLING WHEAT KERNELS USING A BICHROMATIC PULSED LED DESIGN 9.1 INTRODUCTION 9.2 EXPERIMENTAL 12.2.1 WHEAT KERNEL PREPARATION 12.2.2 EXPERIMENTAL APPARATUS AND DESIGN 12.2.3 OPERATION 17.2.4 DATA ANALYSIS 18.3 RESULTS 21.3.1 SIGNAL ACQUISITION 21.3.2 ANGLE AND ANALYSIS METHODS 24.3.3 CONTRAST OF KERNELS 30.3.4 METHOD OF PAIRING LED RESPONSES 31.4 DISCUSSION 33.5 SUMMARY AND CONCLUSIONS 37HAPTER 3. DETERMINATION OF WHEAT KERNEL BLACK POINT DAMAGE USING HYPER-SPECTRAL IMAGING 38.1 INTRODUCTION 38.2 MATERIALS AND METHODS 40.2.1 WHEAT 40.2.2 KERNEL PREPARATION 41.2.3 HYPERSPECTRAL IMAGING SYSTEM 43.2.4 LIGHTING AND CALIBRATION OF REFLECTANCE 45.2.5 LIGHTING AND CALIBRATION OF FLUORESCENCE 46.2.6 KERNEL SPECTRAL FEATURE ANALYSIS 47.2.7 IMAGE PROCESSING 50.2.7.1 SEGMENTATION AND THE KERNEL MASK EXTRACTION 52.2.7.2 EROSION FOR NOISE ELIMINATION AND DILATION FOR KERNEL RECOVERY 52.2.7.3 KERNEL IMAGE FEATURE CALCULATION 52.2.8 STATISTICAL CLASSIFICATION 54.3 RESULTS AND DISCUSSION 55.3.1 LDA ANALYSIS 56.3.2 DISCUSSION 66.4 CONCLUSION 67HAPTER 4. RFID-INTEGRATED MULTI-FUNCTIONAL REMOTE SENSING SYSTEM FOR SEEDLING PRODUCTION MANAGEMENT IN GREENHOUSE 68.1 INTRODUCTION 68.2 MATERIALS AND METHODS 70.2.1 HARDWARE SYSTEM 70.2.1.1 IMAGING SUB-SYSTEM 71.2.1.2 ENVIRONMENTAL FACTORS MEASUREMENT SYSTEM 72.2.1.3 RFID INFORMATION SYSTEM 73.2.2 SOFTWARE PLATFORM 73.2.2.1 IMAGE PROCESSING 74.2.2.1.1 SPATIAL CALIBRATION 74.2.2.1.2 IMAGE STITCHING 74.2.2.1.3 IMAGE SEGMENTATION 75.2.2.2 MIDDLEWARE OF RFID SYSTEM 76.2.3 SAMPLE AND EXPERIMENT GREENHOUSE 76.3 RESULTS AND DISCUSSION 76.3.1 REMOTE SENSING OF CANOPY AND ENVIRONMENTAL FACTOR MEASUREMENTS 76.3.2 RFID-INTEGRATED PRODUCTION TRACEABILITY SYSTEM WITH THE WATERING CRITERIA 81.3.3 IRRIGATION CRITERIA IN THE GREENHOUSE 84.4 DISCUSSION 89.5 CONCLUSION 89HAPTER 5. GNEREAL CONCLUSIONS 91.1 GENERAL DISCUSSION 91.2 RECOMMENDATIONS FOR FUTURE RESEARCH 93EFERENCES 94application/pdf1938597 bytesapplication/pdfen-US生物材料光學檢測麥粒鐮胞菌枯萎症種苗溫室無線射頻辨識系統光譜影像BiomaterialsOptical InspectionWheat KernelFusarium Head BlightSeedlingsGreenhouseRFID (Radio Frequency Identification)Spectral Imagingthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/180248/1/ntu-98-F90631003-1.pdf