臺灣博碩士論文加值系統

近幾年我們生活的時代科技日益進步，在多媒體影像處理這一領域所具備的技術已經不是從前所能夠比擬的，但是還是存在些許問題是值得我們更進一步去探討與改善。我們會把令人印象深刻的景色拍攝並保存下來，當作自己難以忘懷的回憶，但是拍攝的物件是處在移動中或者拍攝者處於晃動的情況下，所拍攝出來的影像容易產生模糊，而這種類型的模糊被稱為運動模糊。在還原模糊影像的過程中，需要一些重要的參數，因此我們希望建立一個有效率的方式可以取得重要的參數。為了貼近人們心中所拍出的影像，眾多學者持續的關注這些問題並努力改善之。
在本論文中，我們利用感測器所收集到的數據並藉由類神經網路訓練來取得較好的分群結果，再透過類神經網路的輸出結果提供使用者重要的參數來降低還原影像所需要的時間。

In recent years, the technology of our life has grown progressively. The technology in the field of multimedia image processing is not comparable to that of the past, but there are still some problems that deserve to be further explored and improved. We will take and preserve the impressive scenery as an unforgettable memory. However, if the object is moving or the photographer is shaking, the captured image is easily blurred, and this blur is called motion blur. In the process of de-blur, some important parameters are needed, so we hope to establish an efficient way to obtain important parameters. In order to get close to the images taken in people's minds, many scholars continue to pay attention to these problems and work hard to improve them.
In this paper, we use the data collected by the sensor and use neural network training to achieve better clustering results and the output from the neural network provides important information to the user to reduce the time required to restore the image.

目錄
摘要 I
Abstract II
誌謝 III
目錄 IV
表目錄 VI
圖目錄 VII
第一章 緒論 1
1.1 研究背景 1
1.2 研究動機 1
1.3 研究目的 2
1.4 論文架構 2
第二章 文獻探討 3
2.1 類神經網路 3
2.1.1 倒傳遞神經網路 (Back Propagation Network, BPN) 4
2.2 點擴散函數 (Point Spread Function, PSF) 6
2.3 慣性感測元件 (Inertial Measurement Unit, IMU) 7
2.4 還原模糊影像方法 7
2.4.1 遞迴法-Lucy-Richardson去模糊 8
2.4.2 非遞迴法-維納濾波去模糊(Wiener Filter Deblurring) 8
第三章 研究架構與方法 10
3.1 收集數據 10
3.2 類神經網路訓練 11
3.3 還原模糊影像 12
第四章 實驗結果 13
4.1 實驗設備 13
4.2 感測器數據收集 14
4.3 影像來源 14
4.4 實驗流程 15
4.5 實驗結果 15
第五章 結論與未來研究 28
參考文獻 29






表目錄
表 1、分群方式。 12
表 2、四種分群差異。 12
表 3、A分群TOP 15：BPN網路T=1,R=50求平均值。 19
表 4、B分群TOP 15：BPN網路T=1,R=50求平均值。 19
表 5、C分群TOP 15：BPN網路T=1,R=50求平均值。 20
表 6、D分群TOP 15：BPN網路T=1,R=50求平均值。 20
表 7、A分群TOP 15：BPN網路T=10,R=50求平均值。 21
表 8、B分群TOP 15：BPN網路T=10,R=50求平均值。 22
表 9、C分群TOP 15：BPN網路T=10,R=50求平均值。 22
表 10、D分群TOP 15：BPN網路T=10,R=50求平均值。 23







圖目錄
圖 1、倒傳遞神經網路架構。 4
圖 2、模糊影像產生流程。 6
圖 3、角度與長度計算示意圖，(a)PSF (b)坐標。 7
圖 4、實驗架構流程圖。 10
圖 5、實驗環境。 11
圖 6、實驗設備(a)GoPro攝影機 (b)線性滑軌 (c)Arduino感測器。 13
圖 7、拍攝影像為0度。 14
圖 8、拍攝影像為30度。 14
圖 9、拍攝影像為90度。 15
圖 10、BPN類神經網路程式碼1。 17
圖 11、BPN類神經網路程式碼2。 18
圖 12、模糊影像1。 24
圖 13、還原影像1 L=10,θ=10。 24
圖 14、模糊影像2。 24
圖 15、還原影像2 L=10, θ=5。 24
圖 16、模糊影像3。 24
圖 17、還原影像3 L=280, θ=0。 24
圖 18、模糊影像4。 25
圖 19、還原影像4 L=100,θ=90。 25
圖 20、模糊影像5。 25
圖 21、還原影像5 L=290,θ=15。 25
圖 22、模糊影像6。 25
圖 23、還原影像6 L=100,θ=89。 25

[1] A. Kumar(2017), “Deblurring of motion blurred images using histogram of oriented gradients and geometric moments,” Signal Processing: Image Communication, Vol. 55, pp. 55-65.
[2] H. Xuan(2011), “Research on quantum adaptive resonance theory neural network,” International Conference on Electronic & Mechanical Engineering and Information Technology, Vol. 8, pp. 3885-3888.
[3] J. M. Lee, J. H. Lee, K. T. Park, and Y. S. Moon(2013), “Image deblurring based on the estimation of PSF parameters and the post-processing,” Optik - International Journal for Light and Electron Optics, Vol.124, No. 15, pp. 2224-2224.
[4] L. A, W. Y, D. F, and W. T. Freeman(2011), “Efficient marginal likelihood optimization in blind deconvolution,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR). pp. 2657-2664.
[5] M. Hagiwara(1996), “Self-organizing feature map with a momentum term,” Neurocomputing, Vol. 10, No. 1, pp. 71-81.
[6] M. k. Alsmadi, K. B. Omar, S. A. Noah, and I. Almarashdah(2009), “Performance comparison of multi-layer perceptron (Back Propagation, Delta Rule and Perceptron) algorithms in Neural Networks,” IEEE International Advance Computing Conference, pp. 296-299.
[7] O. Whyte, J. Sivic, and A. Zisserman(2014), “Deblurring shaken and partially saturated images,” International journal of computer vision, Vol. 110, No. 2, pp. 185-201.
[8] R. Dash, and B. Majhi(2014), “Motion blur parameters estimation for image restoration,” Optik - International Journal for Light and Electron Optics, Vol. 125, No. 5, pp. 1634-1640.
[9] S. Schuon, and K. Diepold(2009), “Comparison of motion de-blur algorithms and real world deployment,” Acta Astronautica, Vol. 64, No. 11-12, pp. 1050-1065.
[10] S. Du, Z. Chen, Z. Yuan, and X. Zhang(2005), “Sensitivity to noise in bidirectional associative memory (BAM),” IEEE Transactions on Neural Networks, Vol. 16, No. 4, pp. 887-898.
[11] V. Kůrková, and M. Sanguineti(2017), “Probabilistic lower bounds for approximation by shallow perceptron networks,” Neural Networks, Vol. 91, pp. 34-41.
[12] A. S. Saleh, N. V. Kalyankar, and S. D. Khamitkar(2010), “Deblured gaussian blurred images,” Journal of Computing, Vol. 2, No. 4, pp.33-35.
[13] H. Simon(2004), Neural Networks and Learning Machines. Prentice Hall International, Inc., third edition.
[14] H. Petar(2013), “Prediction of e-Learning efficiency by neural networks,” Cybernetics and Information Technologies, Vol. 12, No. 4, pp. 98-108.
[15]J. J. Hopfield(1982), “Neural networks and physical systems with emergent collective computational abilities,” National Academy of Sciences, Vol. 79, pp. 2554-2558.
[16] V. Gabriel, F. P. Juan, C. Pablo, and D. P. Fernando(2018), “Artificial neural networks used in optimization problems,” Neurocomputing, Vol. 272, pp. 10-16.
[17] L. David, M. Christian, D. Rene, N. Daniel, and A. Etienne(2007), “A new algorithm for point spread function subtraction in high-contrast imaging: a demonstration with angular differential imaging,” The Astrophysical Journal, Vol. 660, No. 1, pp. 770-780.
[18] F. R. Nuno, and P. S. Cristina(2017), “Inertial measurement units: A brief state of the art on gait analysis,” IEEE 5th Portuguese Meeting on Bioengineering (ENBENG).
[19] A. Kim, and M. F. Golnaraghi(2004), “Initial calibration of an inertial measurement unit using an optical position tracking system,” In Proc. PLANS, IEEE Position Location and Navigation Symposium, pp. 96-101.
[20] S. Isaac, and H. Peter(2006), “Calibration of a mems inertial measurement unit,” In Proc. XVII IMEKO world congress.
[21] Y. Bando(2009), “Single-shot image deblurring with modified camera optics,” PhD Thesis, The University of Tokyo.
[22] S. Kumar, P. Kumar, M. Gupta, and K. N. Ashok(2010), “Performance comparison of median and wiener filter in image de-noising,” International Journal of Computer Applications, Vol. 12, No. 4, pp. 27-31.
[23] S. Swati, S. Shipra, and M. Rajesh(2013), “Image restoration using modified lucy richardson algorithm in the presence of gaussian and motion blur,” Advance in Electronic and Electric Engineering, Vol. 3, No. 8, pp. 1063-1070.
[24] M. Hajmeer, and I. Basheer(2002), “A probabilistic neural network approach for modeling and classification of bacterial growth/no-growth data,” Journal of Microbiological Methods, Vol. 51, pp. 217-226.
[25] B. Shishir, and K. V. Ganesh(2008), “Recognition of facial expressions using Gabor wavelets and learning vector quantization,” Engineering Applications of Artificial Intelligence, Vol. 21, No. 7, pp. 1056-1064.