超音波影像(Ultrasound Image)在醫學影像工具中常扮演著眼睛的角色，可以幫助我們觀察眼睛看不到的身體病變;它具有影像擷取容易、非侵入式、影像及時成像等等優勢，已廣泛應用於各種醫學治療中。另一方面，超音波系統也有其缺點，因為擷取設備的關係，醫療超音波影像有別於一般自然影像處理，其成像結果較不清楚。然而，在醫療超音波影像上，細節資訊表示著可能發生的潛在病變與不尋常狀況，故需要保留以及增強超音波影像中細節紋理以利於醫師判斷病情。本研究提出以Tetrolet轉換，利用其能量集中的特性來保留細節紋理資訊，並加入自適性直方圖等化(Adaptive Histogram Equalization, AHE)，來自動調整超音波影像的對比以達到上述功能。但Tetrolet轉換在VLSI硬體實現時，會有較長的運算複雜度(Complexity)與記憶體(Memory)使用量較大的問題。有鑑於此，本研究提出針對Tetrolet轉換的改良，以降低比對搜尋模板法的速度與記憶體使用量，我們稱之為低複雜度比對模板法(Low-complexity Matching Pattern, LMP)。再利用所提出的LMP並配合查表法(Look-up Table)，來解決AHE在線性內插時所產生硬體運算量過大的問題。由實驗結果得知，改良式架構在運算時間與硬體的提升上，可適合應用於手持式超音波影像系統。最後，我們將其硬體架構實現在Xilinx 7系列的ZedBoard Zynq-7000 ARM/FPGA SoC Development Board上。

An ultrasound image plays the role of eyes in medical imaging tools, and it can help us to observe the physical lesions. The ultrasound image system captures the image more easily than other medical system, and non-invasive detection can reduce the diagnostic time. It is widely used in a variety of medical treatments and medical testing currently. On the other hand, medical ultrasound images are often unclear due to the capturing devices used. As the medical ultrasound images can provide information regarding potential pathological changes and abnormalities, the images must be preserved and enhanced to help doctors to monitor the course of a disease. This research work proposes to use the energy concentration characteristics of Tetrolet transform to preserve the texture information and integrate the technique of adaptive histogram equalization (AHE) to automatically adjust the contrast of an ultrasound image, thereby to resolve the aforementioned function. However, the implementation of the Tetrolet transform in very large scale integration (VLSI) architecture brings issues such as computational complexity and large memory use. In view of these problems, this research work proposes a low-complexity matching pattern (LMP) to reduce the amount of memory used. The LMP is combined with a look-up table approach to solve the issues of the excessive calculations generated in AHE during linear interpolation. The experimental results indicate that the modified algorithm is applicable in hand-held ultrasound imaging systems as it is able to improve calculation times and hardware. Finally, It was implemented in an Xilinx 7 series ZedBoard Zynq-7000 ARM / FPGA SoC Development Board.

目錄

第一章	緒論	1
1.1研究動機與目的	2
1.2論文架構	4
第二章 認識超音波影像	5
2.1 音波基本物理現象	5
2.2 超音波的掃描方式	6
2.3 超音波的顯像模式	7
2.3.1  模式原理:	7
2.3.2  模式原理:	8
2.3.3  模式原理:	9
2.3.4  模式原理:	10
2.4 超音波的雜訊	11
第三章 影像對比調整文獻回顧	12
3.1 全域性直方圖等化(Global Histogram Equalization)	12
3.2區域性直方圖等化(Local Histogram Equalization)	13
3.3 自適性直方圖等化(Adaptive Histogram Equalization)	15
第四章 離散小波轉換文獻回顧	17
4.1 稀疏矩陣（Sparse Matrix）	17
4.2 HAAR小波轉換	20
4.2.1. HAAR小波分解	21
4.2.2 HAAR小波重建	25
4.3 具有稀疏化效果的方向性小波轉換	27
4.3.1  非自適應(Non-adaptive)的小波系統	28
4.3.2  自適應(Adaptive)影像局部結構的小波系統	30
4.4 Tetrolet轉換文獻	33
第五章 Tetrolet轉換演算法	36
5.1 Tetrolet轉換演算法分解步驟	39
第六章 提出的低複雜度超音波影像系統	43
6.1 提出的改良式Tetrolet演算法	45
6.1.1 改良式低複雜度Tetrolet模板	47
6.1.2  改良式Tetrolet拼圖法(Tetrolet Puzzle Method)	49
6.1.3　改良式Tetrolet搜尋法(Tetrolet Search Method)	51
第七章 實驗結果與比較	59
7.1  Tetrolet影像品質比較	59
7.2  Tetrolet轉換軟體計算時間比較	60
7.3  Tetrolet硬體計算時間比較	63
7.4  Tetrolet硬體資源比較	64
7.5  AHE影像對比度比較	65
第八章 結論	67
第九章 參考文獻	68


 
圖目錄

圖2.1、聲頻範圍	5
圖2.2、(a)一維16陣元線性陣列換能器 (b)二維8×8陣元線性陣列換能器…..	6
圖2.3、A Mode模式:二維圖形轉為一維影像	7
圖2.4、多條線掃瞄	8
圖2.5、B Mode模式:形成二維影像	8
圖2.6、M Mode模式:區域的組織影像	9
圖2.7、D Mode 模式	10
圖3.1、(a)原始灰階影像和其直方圖 (b) GHE結果灰階影像與其直方圖		13
圖3.2、區域直方圖計算的區域視窗	14
圖3.3、(a) 原始X-Ray影像(b)經過LHE後的X-Ray影像	14
圖3.4、(a) 影像分割數塊獨立且重疊的子區塊(b)影像區域被切分為許多區塊，包括中心區域A、邊界區域BkH和BkV、角落區域CkI	16
圖4.1、影像原始像素值	21
圖4.2、第一次水平分割圖	21
圖4.3、水平分割後的數值	22
圖4.4、第一次垂直分割圖	22
圖4.5、三階HAAR小波轉換流程圖	23
圖4.6、影像原圖	24
圖4.7、HAAR小波轉換後的圖	24
圖4.8、原始小波係數	25
圖4.9、第一次垂直回復係數矩陣	25
圖4.10、垂直回復後的數值	26
圖4.11、回復原來圖形的像素值	26
圖4.12、反HAAR小波轉換之結果	26
圖4.13、Curvelet分解頻帶劃分示意圖	28
圖4.14、離散Contourlet轉換濾波器組結構圖	29
圖4.15、離散Contourlet轉換的頻域分解圖	30
圖4.16、(a)在區域內流動的例子，每個箭頭是流動向量  (b)影像在適應二維正方形分割與其幾何流向	30
圖4.17、通過簡單路徑程序找到的路徑與最大	31
圖4.18、具有不同的灰度值且同灰階值區域相連的漸變軌跡	31
圖5.1、五種基本單位模塊(Tetrominoes)	37
圖5.2、二十二種基本模板(Pattern)	37
圖5.3、加入旋轉與翻轉後117種模板(Pattern)	38
圖5.4、原始Tetrolet演算法分解步驟流程圖	39
圖5.5、藉由基本模塊(Tetrominoes)覆蓋所組成的4×4模板範例	42
圖6.1、系統流程圖	45
圖6.2、避免重覆比對運算的問題	47
圖6.3、提出89種改良式	48
圖6.4、提出改良式拼圖法流程圖	49
圖6.5、提出改良式拼圖法	50
圖6.6、提出改良式搜尋查表法流程圖	52
圖6.7、範例一：經過三層搜尋並獲得最佳模板	54
圖6.8、範例二：第二層搜尋以獲得最佳模板	54
圖6.9、如範例三所示第一層搜尋以獲得最佳模板………………………………………………………………………..55
圖6.10、N×N(N=4)區塊效應(Block Effect)	56
圖6.11、雙線性內插法	56
圖6.12、可旋轉式的三種基本型模板	57
圖6.13、本文所提出基本型模板：(a)三種基本型模板、(b)加入與鄰近區域關係	58
圖6.14、AHE流程圖	58
圖7. 1、本文實現結果：(a)輸入超音波影像、(b)改良式Tetrolet轉換之LL子頻帶、(c)經對比增強後反轉換的影像、(d)實現在FPGA結果 		66




 
表目錄

表1、轉換時間與影像品質比較	60
表2、Tetrolet轉換分解(Decomposition)步驟之時間比較…..	62
表3、Tetrolet轉換分解(Decomposition)步驟之時間比較	62
表4、比較Tetrolet轉換分解步驟之時間	63
表5、比較Tetrolet硬體資源	64

參考文獻

[1]	Candes E J and Donoho D L. “New tight frames of curvelets and optimal representations of objects with piecewise C2 singularities,” Communications on Pure and Applied Mathematics, 2004, Vol 57,issue 2,page 219-266.
[2]	Do M N and Vetterli M. “The contourlet transform an efficient directional multiresolution image representation,” IEEE Transactions on Image Processing, 2005, 14(12): 2091-2106.
[3]	Huo C, Zhang R, and Yin D. “An adaptive sparse representation for remote sensing image based on combination of wavelet and adaptive directional filter.” Acoustics Speech and Signal Processing (ICASSP), Dallas, 2010: 990-993.
[4]	Le Pennec E and Mallat S. “Sparse geometric image representations with bandelets.” IEEE Transactions on Image Processing, 2005, 14(4): 423-438.
[5]	Plonka G. “The easy path wavelet transform: a new adaptive wavelet transform for sparse representation of twodimensional data.” Multiscale Modeling and Simulation, 2009, 7(3): 1474-1496.
[6]	Plonka G, Tenorth S, and Rosca D. “A new hybrid method for image approximation using the easy path wavelet transform.” IEEE Transactions on Image Processing, 2011,20(2): 372-381.
[7]	S. Gupta and Y. Kaur, “Review of different local and global contrast enhancement techniques for a digital image,” International Journal of Computer Applications, vol. 100, no. 18, pp. 18-23, 2014.
[8]	H. Zhu, F. H.Y. Chan, F. K. Lam, “Image contrast enhancement by constrained local histogram equalization,” Computer Vision and Image Understanding, vol. 73, no. 2, pp. 281-290, 1999.
[9]	S. M. Pizer, E. P. Amburn, J. D. Austin, R. Cromartie, A. Geselowitz, T. Greer, B. H. Romeny, J. B. Zimmerman, and K. Zuiderveld, “Adaptive histogram equalization and its variations,” Computer Vision, Graphics, and Image Processing, vol. 39, no. 3, pp. 355–368, 1987.
[10]	Z. Wang and J. Tao, “A fast implementation of adaptive histogram equalization,” IEEE International Conference on Signal Processing, vol. 2, 2006.
[11]	A. Sanny, Y.-H. E. Yang, and V. K. Prasanna, “Energy-efficient histogram equalization on FPGA,” IEEE High Performance Extreme Computing Conference, pp. 293–299, 2014.
[12]	J.-Y. Kim, L.-S. Kim, and S.-H. Hwang. “An advanced contrast enhancement using partially overlapped sub-block histogram equalization.” IEEE Transactions on Circuits and Systems for Video Technology, vol. 11, no. 4, pp. 475–484, 2001.
[13]	C.-H. Hsia, J.-S. Chiang, and J.-M. Guo, “Memory-efficient hardware architecture of 2-D dual-mode lifting-based discrete wavelet transform,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 23, no. 4, pp. 671-683, 2013.
[14]	J. Krommweh, “Tetrolet transform: A new adaptive Haar wavelet algorithm for sparse image representation,” Journal of Visual Communication and Image Representation, vol. 21, no. 4, pp. 364-374, 2010.
[15]	Z. Peng, L.-B. Tang, and B.-J. Zhao, “Image spare approximation based on Tetrolet transform,” Systems Engineering and Electronics, vol. 38, no. 11, pp. 2536-2539, 2011.
[16]	Y. Chen, R. Zhang, and D. Yin, “SAR image sparse representation based on Tetrolet packet transform,” Journal of Electronics & Information Technology , vol. 34 , no. 2 , 2012.
[17]	C. Li, J. Sun, and Y Kang, “A new algorithm for image denoising based on Tetrolet transform,” Proceeding of SPIE, vol. 7820, 2010.
[18]	J. Gao, Y. Dong, Y. Shen, and C. Zhang, “Research of image fusion algorithm based on improved Tetrolet transform,” Computer Science, vol. 42, no.5, pp. 320-323, 2015.
[19]	H. Chen and Y. Wang, “Research on image fusion algorithm based on Laplacian pyramid transform,” Laser and Infrared, vol.39, no.4, pp. 439-442, 2009.
[20]	K. L. N. Kishore and G. Prathibha, “Multi-focus medical image fusion using Tetrolet transform based on global thresholding approach,” International Journal of Science and Research, vol. 4, no. 1, pp. 518-522, 2015.
[21]	J. Krommweh and J. Ma, “Tetrolet shrinkage with anisotropic total variation minimization for image approximation,” Signal Processing, vol. 90, no. 8, pp. 2529-2539, 2010.
[22]	C. Li, J. Sun, and Y. Kang, “Adaptive image thresholding denosing based on Tetrolet transform,” Jouranl of Hannan University, vol. 28, no. 4, pp. 348-352, 2010.
[23]	P. Jain and V. Tyagi, “An adaptive edge-preserving image denoising technique using Tetrolet transforms,” The Visual Computer : International Journal of Computer Graphics, vol. 31, no. 5, pp. 657-674, 2015.
[24]	A. A. Abdelouahad, M. Omari, H. Cherifi, B. Alibouch, and M. E. Hassouni,” Tetrolet-based reduced reference image quality assessment approach,” International Conference on Multimedia Computing and System, pp. 52-56, 2014.
[25] https://zh.wikipedia.org/wiki/%E5%8C%BB%E5%AD%A6%E8%B6%85%E5%A3%B0%E6%A3%80%E6%9F%A5