臺灣博碩士論文加值系統

近年來，由於在交通便利的環境中，在車輛行駛存在很多的視覺差異，造成駕駛人於死角上的忽視，造成許多的意外情形，所以行車車側安全眼(Side View，SV)對行車安全上十分重要，利用後視鏡的方式來監看物體接近的距離，其車側安全眼利用後視鏡的偏移量來辨識死角方位，利用偏移的方式提供側方盲區影像來避免行車上的碰撞，改善視角不及所造成的意外情形。
在盲點偵測系統(Blind Spot Monitoring，BSM)能在側方的死角狀況進行監控，透過後視鏡的數位攝影監控汽車盲點區域，其演算法是通過計算相鄰的車道距離和方向潛在危險區域進行辨識警告，其主要利用車輛其保險桿設定盲點接近的距離來避免行車碰撞，在全周影像行車輔助系統(Around View Monitoring，AVM)上提出無死角的環視影像系統來監看周遭物體的狀況來避免碰撞情形，本文將提出改善行車上的死角區域來提供即時盲區影像的辨識，利用環景拼接的方式來呈現物體實際距離，將行駛中的車輛所形成盲點區域，改善側方監控的視角範圍，有效的掌控側方物體接近的距離，透過拼接的影像動態資訊來防止意外的產生，同時能改善視野不及的差異，避免死角內輪差所造成的突發狀況。

Because the transportation has been being more convenient in recent years, there are quite a lot of visual differences which could be generated while a vehicle is driven on the road, such situation may cause the driver to ignore the possible dead space and vehicle-related accidents may thereof occur. As such, a Side View(SV) provided during driving plays an important role for the driving safety, a rear mirror is mostly utilized for monitoring the distance of an approaching object, and the side view takes advantage of the offset amount of the rear mirror for identifying the orientation of the dead space, an offset means is utilized for providing a dead space image defined at the side for preventing the vehicle from colliding, so that the problem of having accidents caused by poor viewing angles can be improved.
The Blind Spot Monitoring (BSM) can be used for monitoring when subjected to a dead space at the side, a digital image capturing provided by the rear mirror is served to monitor the blind spot of the vehicle, the algorithm is to calculate the potential dangerous zone defined by the adjacent vehicle lanes and directions for the purpose of identifying and warning, a bumper of the vehicle is set with a blind spot approaching distance for avoiding the vehicle collision. With the Around View Monitoring (AVM) which provides a panoramic viewing system without dead space for monitoring the statuses of neighboring objects for avoiding the colliding situation, this research is aimed to improve the dead space during driving for providing a timing identification regarding the images at the blind zone, and a panoramic stitching means is provided for presenting the actual distance of a certain object, so that the blind zone formed while the vehicle being on the road and the viewing range for the side monitoring can be improved, the approaching distance of the side object can be effectively controlled, and the dynamic information of the stitched images can prevent the accidents from occurring, meanwhile the differentiation formed due to not having full viewing angles can also be improved, and the emergency due to the radius difference of inner wheels at the dead space can be avoided.

摘要 i
ABSTRACT iii
誌謝 v
目錄 vi
圖目錄 viii
表目錄 xii
第一章 導論 1
1.1簡介 1
1.2研究動機 3
1.3論文架構 4
第二章 相關研究與文獻探討 5
2.1車用偵測的發展 5
2.2盲點警示系統 6
2.3全周影像行車輔助系統 9
2.4車側安全眼 10
2.5用於行車側方環景影像拼接安全系統 11
2.6影像匹配相關技術 12
2.7最近鄰近特徵演算法 14
第三章 研究方法 15
3.1拼接系統架構 15
3.2 Overlapping定義的區域 16
3.3擷取靜態圖像的攝影機 17
3.4環繞拼接的實際構想 19
3.5特徵點圖像匹配流程 21
3.6高斯差分尺度空間構建 24
3.7影像拼接系統設計總結 27
第四章 軟硬體設計架構 29
4.1攝影機硬體規格 29
4.2影像拼接的軟體 30
4.3車載影像特徵匹配應用實例 31
第五章 實驗方法與結果 36
5.1影像調整前處理說明 36
5.1.1影像平滑度 37
5.1.2直方圖均衡化 37
5.1.3 Gamma Curve校正 38
5.1.4影像卷積銳化 38
5.2全景拼接模擬實驗方法 40
5.3影像邊緣化的模擬比較 43
5.4全景拼接的實驗結果 53
5.5客觀方法比較 60
5.6行車安全輔助系統設置構想 63
5.7實體街景拼接全景呈現 64
5.8行車側角拼接全景實現 68
第六章 總結與未來展望 71
參考文獻 72

[1] Y. F. Li, Y. M. Wang, W. Q. Huang, and Z. L. Zhang, “Automatic Image Stitching Using SIFT,” International Conference on Audio, Language and Image Processing(ICALIP2008), China, Jul. 2008, pp. 568-571.
[2] D. G. Lowe, “Object Recognition from Local Scale-Invariant Features,” International Conference on Computer Vision(ICCV), Corfu, Sept. 1999, pp. 1150-1157.
[3] W. B. Fei, Y. H. Chen, C. C. Kao, Y. F. Li, and C. J. Chen, “A Vision-Based CollisionWarning System by Surrounding Vehicles Detection,” KSII Transaction on Internet and Information Systems, vol. 6, no. 4, Apr. 2012.
[4] 林信宏，以專利分析探討車輛全周影像發展趨勢，新新季刊第四十三卷第一期.
[5]高祥鈞, 陳加增, 大型車盲點警示系統，中華民國第十四屆車輛工程學術研討會， 虎尾科技大學，台灣雲林，2009.
[6]W. B. Jang, E. Lee, M. R. Park, and D. W. Seo, “Mono-Camera Based Side Vehikle Detection for Blind Spot Detection Systems,” Seventh International Conference on Ubiquitous and Future Networks(ICUFN), pp. 147-149, Jul. 2015.
[7] 紀皓瀚, 許立佑, 楊志弘, 具複合功能全周影像安全系統，中華民國第十八屆車輛工程學術研討會，屏東科技大學，台灣屏東，2013.
[8]H. K. Chang, C. C. Dong, S. R. Moon, and Y. K. Whoi, “Lane Detection System with Around View Monitoring for Intelligent Vehicle,” International SoC Design Conference (ISOCC), pp. 215-218, Nov. 2013.
[9] J. Kuwana, and M. Itoh, “Dynamic Angling Side-View Mirror for Supporting
Recognition of a Vehicle in the Blind Spot,” International Coference on Control Automation and System(ICCAS), Seoul, Oct. 2008.

[10]H. Bay, A. Ess, T. Tuytelaars, and L. G. Van, “Speeded Up Robust Features,” Computer Vision and Image Understanding(CVIU), vol. 110, no. 3, pp. 346-359, 2008.
[11] W. Lin, J. X. Hong, Hao. Zhang, and L. Lin, “Fast and Effective Method for Video
Mosaic,” Computer Engineering and Applications, vol. 45, no. 24, pp. 173-175, 2009.
[12] P. Luca, G. Giorgio, “K-Nearest Neighbors Directed Synthetic Images Injection,” International Workshop on Image Analysis for Multimedia Interactive Services(WIAMIS), Apr. 2010.
[13] G. F. Zhang, H. Yi, W. F. Chen, and J. Y. Jia, “Multi-Viewpoint Panorama Construction with Wide-Baseline Images,” IEEE Transactions on Image Processing, vol. 25, no. 7, pp. 3099-3111, 2016.
[14] S. S. Abood and A. A. Karim, “Fast Generation of Video Panorama,” Internal Journal of Emerging Trends and Technology in Computer Science, vol. 3, no. 6, pp. 80-84, 2014.
[15]D. G. Lowe, “Distinctive Image Feature from Scale-Invariant Keypoints,” International Journal of Computer Vision, Jan. 2004.
[16]K. Hideaki, K. Hiroto, I. Tohru, M. Hiroshi, F. Yoshiya and I. Norikazu, “Generation of Panoramic Image Based on Planar Image Features Using Side-View in-Vehicle Camera,” International Joint Conference(ICCAS-SICE), pp. 4007-4010, 2009.
[17]L. C. Chiu, T. S. Chang, J. Y. Chen and Y. C. Chang, “Fast SIFT Design for Real-Time Visual Feature Extraction,” IEEE Trasations on Image Processing, vol. 22, no. 8, pp. 3158-3167, Aug. 2013.
[18]影像卷積銳化,
http://monkeycoding.com/?p=536
http://arbu00.blogspot.tw/2016/11/opencv7-2d-convolution-image-filtering.html
[19] Y. Xiong and K. Pulli, “Fast Panorama Stitching on Mobile Devices,” in Proc. Digest of Technical Papers International Coference on Consumer Electronics(ICCE), Jan. 2010.
[20] Y. Xiong and K. Pulli, “Color Matching for High-quality Panoramic Images on Mobile Phones,” IEEE Transactions on Consumer Electronics, vol. 56, no. 4, Nov. 2010, pp. 2592-2600.
[21]G. Y. Tian, D. Gledhill, D. Taylor and D. Clarke, “Colour Correction for Panoramic Imaging,” International Conference on Information Visualisation, 2002, pp.483-488.