本研究以追蹤乒乓球為例，建立一個影像動態追蹤的影像伺服控制系統。此系統可偵測移動目標物的形心位置，並將形心位置傳送至運動控制卡進行目標追蹤。影像處理方面，我們利用FPGA實行HSV色彩空間轉換來凸顯目標物顏色特徵，再使用形態學法將多餘雜訊濾除。得到的結果再使用權重法取得目標物的形心位置。此外，為存取影像處理結果，本系統設計影像選取器和影像整合器將影像存取出。控制方法方面，利用微分估測器估測目標物的速度與加速度。並以此為依據估測目標物下一時刻的位置以便提前追蹤。如此可降低追蹤誤差，達到更好的追蹤效果。整體處理程序包含影像處理、訊號傳遞、旋轉平台位置控制，可在攝影機影像更新頻率(60 fps)內完成。實驗結果顯示，硬體影像處理比軟體影像處理的效果好。移動目標物(乒乓球)可持續被追蹤並保持在影像垂直中心附近。

In this study, a rapid image-processing for a moving-object tracking system based on visual servo control is established and verified with a table-tennis-ball-tracking example. The system can detect the centroid of a moving object and moving object is tracked by transferring centroid to motion controller to complete the tracking task. For image processing, field-programmable-gate-array (FPGA) is utilized. At first, the color characteristic of the moving object can be highlighted via using hue-saturation-value (HSV) space transform and morphology to filters out excess noise. After that, centroid of the moving object is calculated by using weight method. Besides, image selector and image integrator are designed to store the result of image processing. As for control, the velocity and acceleration of moving object are calculated by differential estimator to estimate its next position for tracking it in advance. By this way, the tracking error can be reduced to achieve better tracking result. It is noted that, the whole process including image processing, signal transmission and rotational platform control can be finished within the camera's frame rate (i.e., 60 fps). Experimental results show that, image processing is more effective by using hardware than using software , and the moving object (a table tennis ball) can be tracked and maintained close to the vertical center of the image.

[1] Chengtian Song, Keyong Wang, and Lian Zheng. An image processing system
research on target and rendezvous status identi cation of missile and plane.
2008 Congress on Image and Signal Processing, 4:525 { 527, 2008.
[2] Ping-Min Hsu and Zhen-Wei Zhu. Car trajectory prediction in image proces-
sing and control manners. 2016 IEEE International Conference on Intelligent
Transportation Engineering (ICITE), pages 45 { 49, 2016.
[3] Nan Zhang, Yun shan Chen, and Jian li Wang. Image parallel processing
based on gpu. 2010 2nd International Conference on Advanced Computer
Control, 3:367 { 370, 2010.
[4] Al ecio Pedro Delazari Binotto, Dionisio Doering, Thorsten Stetzelberger, Pa-
trick McVittie, Sergio Zimmermann, and Carlos Eduardo Pereira. A cpu,
gpu, fpga system for x-ray image processing using high-speed scienti c came-
ras. 2013 25th International Symposium on Computer Architecture and High
Performance Computing, pages 113 { 119, 2013.
[5] Yun-Won Choi, Yun-Su Chung, Soo-In Lee, and Suk-Gyu Lee. Rear object
detection method based on optical
ow and vehicle information for moving
vehicle. 2017 Ninth International Conference on Ubiquitous and Future Net-
works (ICUFN), pages 203 { 205, 2017.
[6] Wanli Chen, Ya Jun Yu, and Hongjian Shi. An improvement of edge-adaptive
image scaling algorithm based on sobel operator. 2017 4th International Con-
ference on Information Science and Control Engineering (ICISCE), pages 183
{ 186, 2017.
[7] Hu Jiang-Hua, Qin Lei, and Fu Tian-Qi. Image processing based on mathe-
matical morphology in camou
age. 2013 Seventh International Conference
on Image and Graphics, pages 269 { 272, 2013.
[8] Baoshan Zhang, Shoufeng Tang, Manman Jin, Chaoliang Xu, and Minming
Tong. Research on mine robot positioning based on weighted centroid method.
2018 International Conference on Robots & Intelligent System (ICRIS), pages
17 { 20, 2018.
[9] P.M. Sharkey and D.W. Murray. Delay versus performance of visually gui-
ded systems. In IEE Proceedings-Control Theory Application, pages 436{447,
1996.
[10] G.V. McMurray J.A. Piepmeier and H. Lipkin. Tracking a moving target with
model independent visual servoing: a predictive estimation approach. In Pro-
ceedings. 1998 IEEE International Conference on Robotics and Automation,
pages 2652{2657. IEEE, 1998.
[11] H. Araujo J. P. Barreto, J. Batista. Solution for visual control of motion:
Active tracking applications. In Proceedings of the 8th IEEE Mediterranean
Conference on Control and Automation, 2000.
[12] J.L Lai. A study on visual detection and tracking of moving targets. Master's
thesis, National Cheng Kung University, 2013.
[13] E. Brookner. Tracking and Kalman Filtering Made Easy. New York:John
Wiley and Sons, 1998.
[14] Yuan-Pao Hsu, Hsiao-Chun Miao, and Ching-Chih Tsai. Fpga implemen-
tation of a real-time image tracking system. Proceedings of SICE Annual
Conference 2010, pages 2878 { 2884, 2010.
[15] M. C Hanumantharaju, M. Ravishankar, D. R Rameshbabu, and S. Rama-
chandran. A novel fpga implementation of adaptive color image enhancement
based on hsv color space. International Conference on Electronics Computer
Technology, 2:160 { 163, 2011.
[16] Masahiro Shimasaki and Yukinobu Hoshino. Pipeline labeling algorithm with
parallel calculation of gravity center on fpga. International Conference on Soft
Computing and Intelligent Systems (SCIS) and 17th International Symposium
on Advanced Intelligent Systems (ISIS), pages 86 { 91, 2016.
[17] Ko Appiah, Andrew Hunter, Patrick Dickinson, and Jonathan Owens. A
run-length based connected component algorithm for fpga implementation.
2008 International Conference on Field-Programmable Technology, pages 177
{ 184, 2008.
[18] J.G. ZIEGLER and N.B. NICHOLS. Optimum settings for automatic con-
trollers. ASME Transactions, pages 759{768, 1942.
[19] A.S. McCormack and K.R. Godfrey. Rule-based autotuning based on fre-
quency domain identi cation. IEEE Transactions on Control Systems Techno-
logy, 6:43 { 61, 1998.
[20] S.C. Schneider R.H. Brown and M.G. Mulligan. Analysis of algorithms for
velocity estimation from discrete position versus time data. IEEE Transacti-
ons on Industrial Electronics, 39(1):11{19, 1992.
[21] X.W. Lin. Model-fusion-based precision motion control of linear motors. Mas-
ter's thesis, National Cheng Kung University, 2013.