應用於即時影片之穩定技術

隨著資訊科技的進步，影像穩定技術相關產品在市面上規格有些差異，如有無防震、自動對焦等功能都會造成拍攝的影像有差異。若是加上一些不可避免的因素，如環境或人為等因素，而造成影像模糊不清，因此我們就必須去探討如何克服將影像穩定的問題。為了要解決攝影機振動所造成的視訊振動，本文將利用視訊中影像內容進行分析，造成影像的振動因素可分為兩種:物件的運動與攝影機的運動。第一種是攝影機運動固定，則物件運動是局部性的變化，第二種是攝影機振動與物件運動則是全域性的變化。本文採用全域運動估測(global motion estimation)方法來進行矩陣矯正。首先利用光流法(Optical flow)來偵測變異量大的區域，再搭配SURF(speeded up robust features)提取出影像的特徵點，以便找到兩張影像間的轉換矩陣參數。接著估測視訊頁(frame)的振動行為，本研究將從轉換矩陣參數找到像素間位移的對應關係。然後再針對像素間的位移進行影像全域性矯正，最後將校正資料經卡爾曼濾波(kalman filter)得到預測的矯正值，以使得視訊頁(frame)內的運動達到穩定。為了驗證本文提出的應用於即時影像與影片之穩定技術，本研究首先進行了影像穩定效能的比較，實驗的結果本文方法所得PSNR值比其它方法都高，表示本研究之方法穩定效果較佳。接著採用了六種不同的影片，分別進行振動與穩定視訊頁在x與y方向位移的分析與討論，以及PSNR值的分析，由實驗結果得知本文之影像穩定技術能有效解決振動問題。最後，再進行影片的穩定度之ITF評估實驗分析，從實驗結果明顯顯示本文穩定技術所有得ITF值都達到26dB以上。從實驗結果顯示不論在客觀的數據分析與主觀的視覺評比，皆能顯著地提昇影片資料執行影像品質。

關鍵字

SURF ；穩定；光流法；轉換矩陣；卡爾曼濾波

並列摘要

With the progress of information technology, the specifications of image stabilizing technology vary in the market. Functions like the vibration-proof or automatic focus will make a difference to the quality of a photograph. An image may be blurred due to some unavoidable factors, such as environment and people. Therefore, we must discuss the solution to this problem. To eliminate the visual message vibration caused by the shaking of the camera, this paper analyzed the reasons based on the visual message of images. There are two factors that cause image vibration: the movements of the object and the camera. In the first situation, the camera is fixed while the object moves in a certain way while the second one is when both the camera and the object are moving at the same time. This study performed matrix rectification through global motion estimation. Firstly, optical flow was used to detect the zones with tremendous variability. Then SURF (speeded up robust features) was utilized to extract the features of the image. In this way, the matrix parameters of transfer between the two pictures were acquired. Next, the vibration of the frame was estimated. The correspondence between matrix parameters and pixel shift was found. Then global image adjustment was carried out according to the displacement of the pixels. Finally, the forecast value of rectification was obtained by passing the corrected data through the Kalman filter in order to stabilize the movement in the frame. To test the technology, the first step is to compare the effect of image stabilization. The result showed that the Peak Signal to Noise Ratio (PSNR) value of the technology in this paper was higher than any other methods. This means that the technology was able to achieve good performance. Then, the shifts of both vibrant and stabilized frames in six movies at the directions of X and Y were analyzed and discussed. Their PSNR values were also analyzed. The outcome indicated that the image stabilizing technology served as an effective solution to vibration. Finally, the stability of the picture was analyzed through Integrated Test Facility (ITF) ITF evaluation experiment, which told evidently that all the ITF values of the movies using the technology exceeded 26dB. Both objective data analysis and subjective visual estimation proved that the technology presented in this paper could remarkably improve the performance of the data and the quality of a picture.

並列關鍵字

SURF ； Stabilization ； Optical Flow ； Transformation Matrix ； Kalman Filter

參考文獻

1.[1]K. Sato, S. Ishizuka, A. Nikami, M. Sato, “Control techniques for optical image stabilizing system,” IEEE Transactions on Consumer Electronics, vol.39(3), pp.8–10, 1993.

3.[3]C.H. Chen,Y.L. Kuo, “Real-time Video stabilization based on motion compensation,” Innovative Computing, Information and Control (ICICIC), pp.1495–1498,2009.

4.[4]C. Tomasi,T. Kanade, Shape and motion from image streams under orthography: a factorization method,” International Journal of Computer Vision, pp.137–154,1992.

5.[5]R. Madison, “A fast feature tracker for image sequence analysis,” Department of Electric and Computer Engineering, Carnegie Mellon University, 1993.

6.[6]D. Lowe, “Distinctive image features from scale invariant keypoints,” International Journal of Computer Vision 60, pp.91–110,2004.

國際替代計量

應用於即時影片之穩定技術

全文下載

主題瀏覽