Improved algorithm for gradient vector flow based active contour model using global and local information.

Area of Science:

• Computer Vision
• Image Processing
• Medical Imaging

Background:

• Active contour models are essential for object boundary extraction in digital images.
• Existing gradient vector flow (GVF) methods struggle with targets exhibiting deep concavities, leading to poor convergence.
• Accurate segmentation of complex boundaries is crucial in applications like medical image analysis.

Purpose of the Study:

• To propose an improved active contour model that enhances convergence and accuracy for object boundary extraction.
• To address the limitations of existing GVF methods in handling images with deep concavities.
• To develop a more robust method for segmenting complex boundaries in digital and medical images.

Main Methods:

• Developed a novel algorithm to accurately identify false active contour segments using global GVF forces.
• Introduced a new algorithm for updating the external force field, incorporating local magnetostatic force information.
• Implemented a semi-dynamic external force field that adjusts only when false contours are detected, balancing accuracy and computational cost.

Main Results:

• The improved active contour model demonstrated higher accuracy in determining false contour parts.
• The method effectively approximated complex object boundaries, outperforming existing techniques on irregular shapes.
• Experimental validation on MRI and ultrasound medical data confirmed the method's efficiency and robustness.

Conclusions:

• The proposed active contour model significantly improves boundary extraction accuracy, particularly for challenging shapes with deep concavities.
• The integration of GVF and magnetostatic forces, along with a semi-dynamic force field, offers an effective solution for complex image segmentation.
• The method provides an efficient and computationally viable approach for applications in medical image analysis and beyond.