視訊穩定是一個重要的影像增強技術，主要用於移除惱人的影片晃動。視訊穩的問題是:利用軟體來預測畫面的晃動量，並將畫面調整回穩定的狀況。視訊畫面的晃動使人類感受不舒服的狀態，因此視訊晃動畫面的穩定，不僅是用於即時狀況，對於所儲存的資料，亦是非常的重要。因此，發展一快速，又能有效解決畫面晃動的方法是非常必要的。對監錄系統而言，我們需要發展一即時演算法以穩定定點攝影機所取得的資訊，且因為視訊包含保貴的資訊，因此也必須儘量補償畫面穩定後所造成的空白區域。
論文提出的方法包括三個步驟:第一是取得絕對背景，並從許多特徵區塊挑選出一偵測區塊，我們利用邊緣偵測和影像直方圖來求得此偵測區塊。第二是預測攝影機的晃動量，我們利用偵測區塊的比對，求得最小的平均平方差(Mean of square difference)，其發生的位置即是晃動向量，這是一個簡單且快速的晃動向量取得方法。第三是補償背景或是移動物的空白區塊，我們使用絕對背景與FMM(Fast marching method)的方法來補償，此法並不會因為畫面晃動，而造成畫面的模糊。
論文提出的方法經過許多實地資料的測試，其所測得的晃動量是準確的，它使畫面因晃動造成的模糊更清析，它滿足一監錄系統即時處理的要求，而且對於畫面劇烈晃動的情況，它的補償效果亦非常良好。

Video stabilization is an important video enhancement technology which aims at removing annoying shake motions from video. A major problem of current software video stabilization is to estimate the camera motion and then stabilize the frame by using only software techniques. Frame vibration is not comfortable for human visual perception and video stabilization is very important for recorded data to be utilized for computer vision application. Therefore, it is necessary to develop an effective video stabilization method for a specific situation. In a vehicle surveillance system, we need to develop a real-time algorithm for stationary camera to stabilize the captured video. Furthermore, the information of video should not be sacrificed. In other words, we not only stabilize the frame of video but also have to compensate the missing area as could as possible.
Our method follows three steps to accomplish the purpose. The first step is to capture the absolute background (ABG) and select the detection block from candidate feature blocks. Edge detection and pixel value histogram are used to pick out the feature blocks from ABG. The detection block is then selected from the feature blocks. The second step is to estimate the camera motion. The vibration vector of camera motion is obtained by miniming the MSD. It is a simple and efficient way to estimate the camera motion. The third step is to repair the missing area of the background and the moving objects. We use the absolute background and fast marching method to repair the unstable image. It guarantees that the missing area of the image will not be blurred by camera vibration.
The proposed algorithms of block matching and frame repair are tested on many video sequences. The estimated vibration vectors are shown accurate. The deblurred results are also shown for visual comparison. The computational complexity of our method satisfies the real-time requirement of a video surveillance system. Our method solves the background problem in video stabilization even when the camera vibrates dramatically.

[1] H. C. Chang, S. H. Lai and K. R. Lu, “A robust real-time video stabilization algorithm,” Journal of Visual Communication and Image Representation, Vol. 17, No. 3, Pages: 659 - 673, June 2006.
[2] Z. Duric and A. Rosenfeld, “Stabilization of image sequences,” Tech. Rep. CAR-TR-778, July 1995.
[3] S. Etrurk, “Digital image stabilization with sub-image phase correlation based global motion estimation,” IEEE Trans. on Consumer Electronics, Vol. 49, No. 4, Pages: 1320 - 1325, November 2003.
[4] S. Etrurk, “Real-time digital image stabilization using Kalman filters,” Real-Time Imaging, Vol. 8, No. 4, Pages: 317 - 328, August 2002.
[5] J. Flusser and T. Suk, “A moment-based approach to registration of images with affine geometric distortion,” IEEE Tran. on Geoscience and Remote Sensing, Vol. 32, No. 2, Pages: 382 - 387, March 1994.
[6] C. Gonzalez and Woods, Digital Image Processing, 2nd ed, U.S., Prentice Hall, 2002.
[7] A. Goshtasby , G. C. Stockman and C. V. Page , “A region-based approach to digital image registration with subpixel accuracy,” IEEE Trans. on Geoscience and Remote scensing, Vol. GE-24, No. 3, Pages: 390 - 399, May 1986.
[8] C. Guestrin, F. Cozman and E. Krotkov. “Image stabilization for feature tracking and generation of stable video overlays,” Technical Report CMU-RI-TR-97-42, November 1997.
[9] M. Hansen, P. Anandan, K. Dana, G. V. D. Wal and P. Burt, “Real-time scene stabilization and mosaic construction,” Proc. of IEEE on Computer Vision and Pattern Recognition, Pages: 54 - 62, December 1994.
[10] B. K. P. Horn and B. G. Schunck, “Determining optical flow,” Artificial Intelligence, Vol 17, Pages: 185 - 203, 1981.
[11] D. P. Huttenlocher, G. A. Klanderman and W. J. Rucklidge, “Comparing images using the Hausdorff distance,” IEEE Trans. on Pattern Analysis and Machine Intellinence, Vol. 15, No. 9, Pages: 850 - 863, September 1993.
[12] M. Irani and P. Anandan, “Video indexing based on mosaic representations,” Proc. of IEEE, Vol. 86, No. 5, Pages: 905 - 921, May 1998.
[13] J. Jia, T. P. Wu, Y. W. Tai and C. K. Tang, “Video repairing: inference of foreground and background under severe occlusion,” Proc. of IEEE Conf. on Computer Visio and Pattern Recognition, Vol. 1, Pages: 364 - 371, July 2004.
[14] J. S. Jin, Z. Zhu and G. Xu, “A stable vision system for moving vehicles,” IEEE Trans. on Intelligent Transportation Systems, Vol. 1, No. 1, Pages: 32 - 39, March 2000.
[15] A. Litvin, J. Konrad and W. C. Karl, “Probabilistic video stabilization using Kalman filtering and mosaicking,” Proc. of IS&T/SPIE Electronic Imaging, Image and Video Communications and Processing, Vol. 5022, Pages: 663 - 674, January 2003.
[16] B. D. Lucas and T. Kanade, “An iterative image registration technique with an application to stereo vision,” Proc. of DARPA Image Understanding Workshop, Pages: 121 - 130, 1981.
[17] Y. Matsushita, E. Ofek, X. Tang and H. Y. Shum, “Full-frame video stabilization,” Proc. of CVPR, 2005.
[18] C. Morimoto and R. Chellappa, “Fast electronic digital image stabilization,” Tech. Rep. of CVL, 1995.
[19] C. Morimoto and R. Chellappa. “Automatic digital image stabilization,” IEEE Conf. on Pattern Recognition, August 1996.
[20] N. R. Pal and S. K. Pal, "A review on image segmentation techniques," Pattern Recognition, Vol. 26, No. 9, Pages: 1277 - 1294, 1993.
[21] M. Roux, “Automatic registration of SPOT images and digitized maps,” Proc. of IEEE Conf. on Image Processing, Pages: 625 - 628, 1996.
[22] M. Sester, H. Hild and D. Fritsch, “Definition of ground control features for image registration using GIS data,” Proc. of the Symposium on Object Recognition and Scene Classification from Multispectrak and Multisensor Pixels, Vol. 32, Part 3/1, Pages: 537 - 543, 1998.
[23] A. Telea, “An image inpainting technique based on the fast marching method,” Journal of graphics tools, Vol. 9, No. 1, Pages: 25 - 36, 2004.
[24] D. J. Wang, T. H. (C. H.) Chen, Y. C. Chiou and H. S. Liau, “An efficient moving object extraction algorithm for video surveillance,” 9th International Conference on Knowledge-Based Intelligent Information and Engineering Systems, Pages: 346-352, September 2005.
[25] Y. Wexler, E. Shechtman and M. Irani, “Space-time video completion,” Proc. of IEEE Conf. on Computer Vision and Pattern Recognition, Vol. 1, Pages: 120 - 127, July 2004.
[26] Y. S. Yao and R. Chellappa, “Electronic stabilization and feature tracking in long image sequences,” Tech Rep.CAR-TR-790, September 1995.
[27] Q. Zheng and R. Chellappa, “A computational vision approach to image registration,” IEEE Trans. on Image processing, Vol. 2, Pages: 311 - 326, March 1993.
[28] B. Zitova, J. Flusser and F. Sroubek, “Application of image processing for the conservation of the medieval mosaic,” Proc. of IEEE Conf. on Image Processing, Vol. 3, Pages: 993 - 996, June 2002.
[29] B. Zitova and J. Flusser. “Image registration methods: a survey,” Image and Vision Computing, Vol. 21, No. 11, Pages: 977 - 1000, October 2003.
[30] I. Zoghlami, O. Faugeras and R. Deriche, “Using geometric corners to build a 2d mosaic from a set of images,” Proc. of IEEE Conf. on Computer Vision and Pattern Recognition, Pages: 420 - 425, June 1997.