Title Page
Contents
요약 8
I. INTRODUCTION 9
1. Motivation 9
2. Task 10
3. Related Work 11
4. Forest Inventory 13
4.1. Importance of forest inventory 13
5. Proposal for Improvements 15
6. Objectives 16
II. LITERATURE REVIEW 17
1. Image Processing 17
1.1. Remote Sensing 17
1.2. Multispectral sensors imagery vs. imagery from cameras 18
1.3. UAV drone imagery vs satellite imagery 18
2. ITCD on spectral imagery 19
3. ITCD on digital imagery 23
4. Summary 25
III. METHODOLOGY 26
1. Introduction 26
2. Fundamentals of Digital Image Processing 26
2.1. Colour vision 27
2.2. Image Restoration 27
2.3. Image Segmentation 28
2.4. Morphological Image Analysis 30
3. Overview of the methodology 32
4. Image Acquisition 34
5. Study Sites 34
6. Pre-processing 36
6.1. Decorrelation Stretching 36
6.2. Median Filtering 38
6.3. Morphological Opening-by-Reconstruction 39
7. Segmentation 39
7.1. Minimum and Maximum Filtering 39
7.2. Sobel Edge Operator 40
7.3. Binary Thresholding 41
8. Postprocessing 42
8.1. Morphological Postprocessing 42
8.2. Region Filling 42
8.3. Removing Small Noise or Objects 42
8.4. Smoothing the Jagged Edges 43
8.5. Object Extraction 43
9. Summary 44
IV. RESULT 45
1. Introduction 45
2. Experimental Result 45
2.1. Pre-processing 45
2.2. Segmentation 48
2.3. Postprocessing 51
3. Statistical measurement 53
4. Summary 55
V. DISCUSSION 56
1. Introduction 56
2. Result Evaluation 56
2.1. Experimental Assumptions 57
3. Comparison with Previous Researches 58
4. Contributions 62
5. Limitations 63
6. Summary 64
VI. CONCLUSION AND FUTURE WORK 65
1. Conclusion 65
2. Future project 65
VII. REFERENCE 67
VIII. APPENDIX 69
Abstract 75
Table 4.1. The statistical measurement for image taken at 50 meter. 53
Table 4.2. The statistical measurement for image taken at 100 meter. 54
Table 4.3. The result evaluation of the algorithm. 55
Table 5.1. The previous studies on digital image processing. 59
Picture 1.1. Forest Management Plan 9
Picture 2.1. Electromagnetic Spectrum. 17
Picture 3.1. Basic flow diagram of the steps of image processing. 26
Picture 3.2. (a) Original Image (b) Thresholded Image for object of interest. 29
Picture 3.3. Block diagram of the tree detection method. 32
Picture 3.4. DJI Phantom 4 Pro. 34
Picture 3.5. Study site from FRIM. 35
Picture 3.6. Study site from Gua Musang. 35
Picture 3.7. Study site from HCVF. 36
Picture 3.8. Eight-neighbourhood 43
Picture 3.9. Connected Component Labelling 44
Picture 4.1. Original image from image acquisition. 45
Picture 4.2. Contrast stretching 46
Picture 4.3. Decorrelation stretching. 46
Picture 4.4. Conversion to greyscale followed by median filtering. 47
Picture 4.5. Subsequent steps of morphological opening-by-reconstruction. 47
Picture 4.6. (a) The normal morphological opening operation. (b) The... 48
Picture 4.7. The regional maxima filter(left). 48
Picture 4.8. The addition of both resultant images. 49
Picture 4.9. The detection of the tree crown boundaries using Sobel edge detector. 49
Picture 4.10. The superimpose of Sobel edge detector. 50
Picture 4.11. The segmentation of tree crown boundary using global binary... 50
Picture 4.12. The pixel connectivity for two-dimensional image. 52
Picture 4.13. The binary image from global thresholding method (left) and the... 52
Picture 4.14. Boundary tracing of the tree crown. 53
Picture 5.1. (a) The scatter plot from the original image. 57
Picture 5.2. The results from previous researches. 60
Picture 5.3. (a) The original image used in the testing of the watershed... 61
Picture 5.4. Partial tree crowns that were selected in the result of the algorithm. 64
Picture 6.1. The species matching using Computer Vision technique. 65