|Budget Amount *help
¥3,400,000 (Direct Cost : ¥3,400,000)
Fiscal Year 2003 : ¥1,400,000 (Direct Cost : ¥1,400,000)
Fiscal Year 2002 : ¥2,000,000 (Direct Cost : ¥2,000,000)
Region extraction from images is an important topic in image processing. For this issue, active contour model (ACM) has achieved considerable success.
In region extraction with ACM, a user first draws a contour C around a target region and defines its evaluation (energy) function B. This C is iteratively deformed so as to decreases B, and the deformed C that minimizes E is extracted as a target region boundary. Since naive ACM evaluates only the certainty that C is located on a region boundary, when an image has several object regions, there is no guarantee that C is deformed to correspond with the target region boundary. To solve this problem, several methods have been proposed for employing the a priori knowledge of a target region shape. However, the existing methods need the making of prototypes or training with samples, and thus they require time-consuming preparation work and impose tight restrictions on their target regions.
To overcome the difficulties in the existing methods, we propose a novel method for employing the a priori knowledge of a target region shape. The proposed method is based on an idea that an initial value of C (an initial contour Cini drawn around a target region by a user) can include the shape features of a target region. In this method, the sequence of shape features (such as corner, line, and curve) is obtained from Cini. This sequence is encoded into a symbolic description and used as the a priori knowledge of a target region shape. During region extraction procedure, by finding the optimal assignment of the symbols (shape features) to the control points of C and determining the optimal positions for these points, the a priori knowledge is reflected on the extracted region shape.