動態攝影之模糊影像還原研究

Deblurring process of blurred image for dynamic photography

邱智勇(ZHI-YONG QIU)

指導教授 ： 李朱慧

朝陽科技大學/資訊學院/資訊管理系/碩士(2016年)

摘要

在現今影像技術的蓬勃創新，已發展出許多的多媒體影像設備，例如行車紀錄器、智慧型手機、運動攝影機、單眼相機…等等。雖然這些多媒體影像設備中影像技術已有一定的水準，但仍存在著一些有關於影像分析上潛在地問題，例如在高速移動的物體上，嘗試拍攝動態影像，容易造成影像中的主體模糊；另外在昏暗或明亮的場景中拍攝主體，容易造成影像曝光的過多或過少。這些潛在地挑戰都值得我們來加以研究及改善。在本論文，我們將探討運動攝影機中影響動態模糊影像的原因，以及嘗試將動態模糊的影像進行去模糊處理。在實驗的過程中，我們將攝影設備架設在線性滑動軌道上來拍攝不同速度與加速度的動態影像，並結合慣性感測元件（Inertial Measurement Unit, IMU）來收集動態影像的三軸加速度數據，便於實驗下階段的分析，預先將三軸加速度數據進行卡爾曼濾波處理。以及根據在不同移動速度與加速度資料所得到的模糊影像，我們將建置在不同移動速度的情況下，每組有效性的點擴散函數的參數系統，而我們使用維納濾波將動態模糊影像進行去模糊處理。

關鍵字

卡爾曼濾波 ； 動態模糊 ； 維納濾波 ； 點擴散函數

並列摘要

In recent years, imaging technology has been flourishing innovation. There have developed a number of multimedia imaging equipment, such as vehicle driving recorders, smartphone photography, sport cameras, SLR cameras etc. Although the technic in these multimedia imaging equipment reaches a certain level, but there are still some potential issue of the image analysis. Such as in the high speed moving objects, attempting to capture motion pictures, the objects likely to cause the image blur. In addition, if capture objects in dark or bright scenes, the image is likely to underexposure or overexposure. These potential challenges are worth to be studied and improved. In this paper, we will explore the influence of the dynamic camera motion blur reason and attempt to perform de-blurring on motion blurred image. During the experiment, we will set up the photographic equipment to record motion images at different speeds on sliding track, and combined X-IMU (Inertial Measurement Unit) to collect data of the triaxial accelerometer motion. In order to facilitate analysis of the next stage of the experiment, We use Kalman filtering to deal for noise reduction with data of the triaxial accelerometer motion. According to capture motion images at different speeds and acceleration data, building the motion image de-blurring system based on the effectiveness parameters of point spread function.

並列關鍵字

Kalman filter ； Motion blur ； Wiener filter ； Point spread function

參考文獻

[1] Ayers, G.R., and Dainty, J.C. (1988), “Interative blind deconvolution method and its applications,” Opt. Lett

[2] Biemond, J., Lagendijk, R., and Mersereau, R. (1990), “Iterative Methods for Image Deblurring,” In Proc. IEEE, Vol. 78, pp. 856–883.

[4] Couzinie-Devy F., Sun J., Alahari K., and Ponce J. (2013), “Learning to estimate and remove non-uniform image blur,” IEEE Conf., In CVPR, pp. 1075–1082.

[7] Eltoukhy, H.A., and Kavusi, S. (2003), “A computationally efficient algorithm for multi-focus image reconstruction,” In Proc. SPIE Electron. Imaging, Vol. 5017, pp. 332–341.

[8] Faragher, R. (2012), “Understanding the basis of the Kalman filter via a simple and intuitive derivation,” IEEE Signal Process. Mag., Vol. 29, No. 5, pp. 128–132, Sep.