課程名稱︰電腦視覺 課程性質︰資工系選修 課程教師：傅楸善 開課學院：電機資訊學院 開課系所︰資訊工程學系 考試日期（年月日）︰2024/10/22 考試時限（分鐘）：180 分鐘 試題 : 1. (62%) Please define the following terms and explain the content, purpose, and application of each term and give an illustrative example if possible. If possi- ble, define the term in mathematical equation. For example: thresholding: an image point operation that produces a binary image from a gray scale image. A binary-1 is produced on the output image whenever a pixel value on the input image is above a specified minimum threshold level. A binary-0 is produced otherwise. Alternatively, thresholding can produce a binary-1 on the output image whenever a pixel value on the input image is below a specified ma- ximum threshold level. A binary-0 is produced otherwise. (1) template matching (2) feature extraction (3) region centroid (4) region area (5) image segmentation (6) edge linking (7) corner finding (8) labeling (9) noise suppression (10) background normalization (11) 2D discrete Euclidean space (12) 4-connectivity (13) border pixel (14) region perimeter (15) octagon (16) GLCM: Gray Level Co-occurrence Matrix (17) image contrast (18) Bayes decision rule (19) economic gain matrix (20) statistical pattern recognition (21) conditional probability (22) fair game assumption (23) reserving judgment (24) maximin decision rule (25) dilation (26) erosion (27) opening (28) closing (29) mathematical morphology (30) operator extensive (31) idempotent 2. (8%) When binarizing image, we should find the best threshold of this image. Therefore, please list two ways (statistical methods) to find the best threshold of the image and explain advantages and disadvantages in detail. 3. (6%) In the Classical Connected Components Labeling Algorithm, we face a big problem - global equivalence table may be too large for memory, so what kind of methods can solve this problem? Please explain differences, advantages, and dis- advantages. 4. (6%) Please describe the method, steps, and results of YouAnamoly: Unified Anamoly Detection for Wafer Defect Inspection. 5. (6%) Please describe the method, steps, and results of Fast In-Bed Human Pose Estimation Using RGB-D (Red, Green, Blue -- Depth) Images. 6. (6%) Please describe the method, steps, and results of YuLPR: Moving Vehicle License Plate Recognition. 7. (6%) Please describe the method, steps, and results of Text to Image Genera- tion with Stable Diffusion. --

※ 發信站: 批踢踢實業坊(ptt.cc), 來自: 36.230.36.163 (臺灣) ※ 文章網址: https://webptt.com/m.aspx?n=bbs/NTU-Exam/M.1731406688.A.062.html ※ 編輯: xavier13540 (36.230.36.163 臺灣), 11/12/2024 18:18:36