Volume Catcher


It is difficult to obtain a specific region within unsegmented volume data (region of interest, ROI). The user must first segment the volume, a task which itself involves significant user intervention, and then chooses a desired target within the 3D space. This paper proposes a simple and intuitive user interface for the task: the user traces the contour of the target region using a 2D free form stroke on the screen, and the system instantly returns a plausible 3D region inside the stroke by applying a segmentation algorithm. The main contribution is that the system infers the depth information of the ROI automatically by analyzing the data, whereas existing systems require the user to provide the depth information explicitly. Our system first computes the 3D location of the user-specified 2D stroke based on the assumption that the user traced the silhouette of the ROI, that is, the curve where the gradient is perpendicular to the viewing direction. The system then places constraint points around the 3D stroke to guide the following segmentation. Foreground constraints are placed inside the stroke and background constraints are placed outside the stroke. We currently use the statistical region-merging algorithm of Nock et al. [Nock04] to perform the segmentation. We tested our system with real-world examples to verify the effectiveness of our approach.

Shigeru Owada, Frank Nielsen, and Takeo Igarashi, "Volume Catcher", Proc. ACM Symposium on Interactive 3D Graphics and Games 2005 (To Appear)
PDF (607KB) Video (wmv,4.4MB)

Source code
#The corresponding patent was submitted by Sony CSL