Enhanced 2-Dimensional to 3-Dimensional Conversion for Medical Images
C. Saranya1, G. Padmavathi2
1C. Saranya, M. Phil Research Scholar, Department of Computer Science, Avinashilingam Deemed University for Women, Coimbatore, India.
2Dr. G. Padmavathi, Professor & Head, Department of Computer Science, Avinashilingam Deemed University for Women, Coimbatore, India.
Manuscript received on October 13, 2011. | Revised Manuscript received on October 24, 2011. | Manuscript published on November 05, 2011. | PP: 342-348 | Volume-1 Issue-5, November 2011. | Retrieval Number: E0241101511/2011©BEIESP
Open Access | Ethics and Policies | Cite
© The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: The advent of modern 3D technological devices and the desire to create 3D images from the numerous 2D images have increased tremendously. Image-based 3D-modeling techniques for creating a 3D representation of a scene from one or more 2D images have received great attention and methods to improve the conversion process have been probed by both academicians and researchers. This research study focuses on particular conversion method recommended that facilitates various image processing algorithms to improve the conversion of 2D images to 3D images. The primary steps involved are Motion, Edge detection and image segmentation, Depth estimation and Shift algorithm. A weighted motion detection registration method is used to calculate the difference between the current image frame and the previous image frame. During edge detection, Sobel edge detector is used to detect the edges. The result of motion detection and edge detection are combined together and then a gray level closing is performed to make the edges connected and smooth. An edge registration module is used to store the motion and edge information in the memory. The segmentation process uses two algorithms, namely, K-Means and Mean Shift. An enhanced connected component algorithm which improves the traditional algorithm to use Max-Tree is used to create refined components. The final step of the proposed algorithm is the shift algorithm, which reconstructs the 3D image. To prove the efficiency of the proposed algorithm, several experiments are conducted. Various parameters like Root Mean Square Error (RMSE), Peak Signal to Noise Ratio, Standard Deviation and Speed of conversion are used to analyze the performance and efficiency of the proposed conversion algorithm. The experimental results proved that the depth map generated by mean-shift algorithm and enhanced connected component produce efficient and improved results.
Keywords: RMSE, Dept Estimation, Edge detection, Shift Algorithm, Sobel Edge Detector.