Improved image segmentation method for geometric active contour

Abstract

The invention discloses an improved image segmentation method for a geometric active contour. The method comprises the following steps: Step 1, inputting a to-be-processed image and a vector field; Step 2, sampling the vector field through a rectangular grid, obtaining an observation set after sampling, and embedding an iterative robust estimator to eliminate errors and noise of observation valuesin the observation set; inserting an observation point, and constructing an advection vector field by utilizing the smoothed ridge regression and constraining the elastic net; and Step 3, embedding the advection vector field and the diffusion flow into the geometric activity contour to construct a unified geometric activity contour model improved based on the advection vector field and the diffusion flow to guide updating of an initial level set, and evolving an activity contour curve to an accurate target contour by the updated level set so as to complete segmentation of the image. Comparedwith a traditional scheme, the method provided by the invention has obvious advantages, and a good segmentation effect is obtained.

Description

technical field [0001] The invention relates to an improved image segmentation method of geometric active contours, which belongs to the field of image processing. Background technique [0002] Contours captured from image sequences have rich spatial structures and have been widely used in video surveillance, medical analysis, and action recognition, etc. Features of interest in images, whether rigid or non-rigid, usually involve complex motion and deformation. To accomplish this challenging task, the Active Contours (AC, Active Contours) method starts evolution curves near object boundaries, then matches the evolution curves to the real contour, and finally resides on the contour. Geometric active contours (GAC) capture the position of parametric contours by minimizing a combination of smooth and gradient-driven energies. [0003] The curve evolution of the GAC model is essentially the diffusion regarded as a spatial scalar model, such as a level set function, and the non...

AI Technical Summary

Problems solved by technology

However, given the sampling interval, even if diffusion via Gaussian smoothing has been pre-implemented to the original image, the global motion between adjacent images can be so significant that the DCF is temporally discontinuous

Images