臺灣博碩士論文加值系統

影像切割的定義為根據影像的紋路及內容，分割成數個不重覆的區域。動態輪廓模型（active contour model）是一個很好的影像切割技術，因為它具有計算簡單及效率的特性。但是傳統的模型上有一些難以克服的問題。第一為傳統模型無法去取得物件凹陷的輪廓。第二是影像的背景紋路或雜訊會嚴重的影響結果的正確性。此外模型中權重參數的設定相當困難，要對每個不同影像找出最適合的參數是相當耗時的工作。因此，動態輪廓模型大都只應用於特定的領域，以預先設定的模型參數去解決特殊的圖片問題。在本論文中，我們提供了一個以內容觸發為基礎的適應性動態輪廓模型(content-triggered adaptive snake model)，簡稱為CASM。第一部份，利用一個快速的預測評估函式，找出可能為物件邊界的梯度分佈範圍。當前進中的snaxels落入這個可能範圍時，CASM根據周遭的灰階梯度變化及snake移動時所造成的內部力量變化，對能量函式的權重做調整－同時對內部及外部力量做正規化的調整。在主要收斂結束後，CASM對於遠離物件邊界不合格snaxels做進一步的修正評估。利用block為基礎的紋路分析找出這些不符資格的snaxels，並用一個有效率的評估方式去改變內力的作用方向，使snaxels向內或外朝物件邊緣移動。實驗證明CASM能夠大幅改善傳統動態輪廓模型，在具有雜訊的環境中的表現也優於梯度向量流場(Gradient Vector Flow)的蛇模型。

Image segmentation is defined as partitioning an image into non-overlapping regions based on the intensity or texture. The active contour model (Snake model) is one of the preferred approaches for image segmentation because of their computational efficiency and simplicity. But, these methods have some problems is hard to overcome. The first problem is the concave boundary which traditional snake is impossible to capture the concave boundary of interested image region. The second problem is the background texture and noisy of image significantly influence the segmentation result. Furthermore, the model’s weight parameters are hard to design, find out the suitable parameters for each different image is a exhausted work. So most applications of snake model are domain-specific, while specifying fixed snake coefficients to an image in problem. In this paper, we propose content-triggered adaptive snake model (CASM) to lead all the parameters of snake model to be automatically adapted to various images in the noisy environment. First, the CASM applies a fast estimation method to find the possible ranges of gradient magnitudes of object boundary. As soon as the gradient magnitude of progressing snaxels falls in those ranges, CASM will adapt the weights within the snake forces of these snaxels according to encountered changes in gray levels and influences of various forces in the resided snake segments. And, it simultaneously renormalizes their external and internal forces. After primary convergence, CASM fires a compensation evolution to rectify the unqualified snaxels far from the object boundary. The unqualified snaxels, which are discovered by block-based texture analysis, can be pushed inward or pulled outward to the object border by so-called directional compensation evolution using an efficient alternative for force directions in revived evolutions. The simulation results demonstrate that CASM can improve the performance of snake very much, and outperform Gradient Vector Flow (GVF) in noisy images.

Abstract(Chinese) i
Abstract ii
CONTENTS v
List of Figures vii
Chapter1 Introduction 1
1.1. Background 1
1.2. Problem description 2
1.3. Organization of thesis 5
Chapter2 Related works 6
2.1. Active contour model (Snake) 6
2.2. Balloons Model 8
2.3. Gradient Vector Flow Snake 9
Chapter3 Ranging likely Boundary Gradient Magnitudes 11
3.1. Prediction of border gradient magnitude 11
3.2. Statistical decision of immersion step and border zone 14
Chapter4 Adaptation of Active Contour Coefficients 18
4.1. Adapting weighting factors of tensile and flexural force 22
4.2. Re-normalization between the internal and external forces 23
4.3. Modifying damping factors 24
Chapter5 Refinement and Rectification using Textural Features 26
5.1. A. Discrimination of the unqualified snake segments 26
5.2. Directional Compensation Evolution 30
Chapter6 Experiment results 33
Chapter7 Conclusions and Future Works 39
REFERENCE 40

[1] Shannon C. E., Weaver W., “The Mathematical Theory of Communication”, Urbana: University ofIllinois Press, 1949.

[2] Chien S. Y., Huang Y. W., B. Y. Hsieh, S. Y. Ma and L. G. Chen, “Fast Video Segmentation Algorithm With Shadow Cancellation, Global Motion Compensation, and Adaptive Threshold Techniques,” IEEE Transactions on Multimedia, VOL. 6, NO. 5, pp. 732-748, OCTOBER 2004.

[3] Cohen L. D., “On Active Contour Models and Balloons,” In CVGIP: Image Understanding, VOL. 53(2), pp. 211-218, MARCH 1991.

[4] Correia P. Pereira L., F., “Objective Evaluation of Video Segmentation Quality,” IEEE Transactions on Image Processing, VOL. 12, NO. 2, pp. 186-200, FEBRUARY 2003.

[5] Davies R. H., Twining C. J., T. F. Cootes, J. C. Waterton and C. J. Taylor, “A Minimum Description Length Approach to Statistical Shape Modeling,” IEEE Transactions on Medical Imaging, VOL. 21, NO. 5, pp. 525-537, MAY 2002.

[6] Gao H., Siu W. C. and Hou C. H., “Improved Techniques for Automatic Image Segmentation,” IEEE Transactions on Circuits and Systems for Video Technology, VOL. 11, NO. 12, pp. 1273-1280, DECEMBER 2001.

[7] Kass M., Witkin A. and D. Terzopoulos, “Snakes: Active Contour Models,” International Journal of Computer Vision., VOL. 1, pp. 321-331, 1987.

[8] Leung C. K., and Lam F. K., “Maximum Segmented Image Information Thresholding,” Graphical Models and Image Processing, Vol. 60, pp.57-76, 1998.

[9] Liu J., Yang Y. H., “Multiresolution Color Image Segmentation,” IEEE Transactions on Pattern Analysis Machine Intelligence, VOL. 16, NO. 7, pp. 689-700, JULY 1994.
[10] Mclnerney T. and Terzopoulos D., “ T-snakes: Topology adaptive snakes” Medical Image Analysis, VOL. 4 pp.73-90, 2000.

[11] Murray D., Basu A., “Motion Tracking with an Active Camera,” IEEE Transactions on Pattern Analysis Machine Intelligence, VOL. 16, NO. 5, pp. 449-459, MAY 1994.

[12] Neumann A., “Graphical Gaussian Shape Models and Their Application to Image Segmentation,” IEEE Transactions on Pattern Analysis Machine Intelligence, VOL. 25, NO. 3, pp. 316-329, MARCH 2003.

[13] Park H., Schoepflin T. and Kim Y., “Active Contour Model with Gradient Directional Information: Directional Snake,” IEEE Transactions on Circuits and Systems for Video Technology, VOL. 11, NO. 2, pp. 252-256, FEBRUARY 2001.

[14] Uchiyama T., Arbib M. A., “Color Image Segmentation Using Competitive Learning,” IEEE Transactions on Pattern Analysis Machine Intelligence, VOL. 16, NO. 12, pp. 1197-1206, DECEMBER 1994.

[15] Xu C. and Prince J. L., “Snakes, Shapes, and Gradient Vector Flow,” IEEE Transactions on Image Processing, VOL. 7, NO.3, pp. 359-369, MARCH 1998.

[16] Xu J. B., L. M. Po and Cheung C. K., “Adaptive Motion Tracking Block Matching Algorithm for Video Coding,” IEEE Transactions on Circuits and Systems for Video Technology, VOL. 9, NO. 7, pp. 1025-1029, OCTOBER 1999.