Interactive image segmentation correcting method based on geodesic active region models

Abstract

The invention discloses an image segmentation correcting method based on geodesic active region models, mainly solving the problem of incorrect probability modeling of a traditional method and comprising the realizing steps of: (1) carrying out Gabor filtering on images to be segmented and extracting the sliding window energy characteristics of each sub-band of the filtered images; (2) establishing Gaussian models according to the sliding window energy characteristics of marking points and estimating the self-adaptive weight of each characteristic passage; (3) adaptively weighting each characteristic passage probability to obtain the region probability and the boundary probability of pixels; (4) correcting the region probability and the boundary probability of the pixels according to a correcting algorithm based on the marking points; and (5) continuously evolving an initial curve by the geodesic active region models to obtain the boundary of an interesting target for finishing segmenting the images. The method enhances the modeling correctness of the region probability and the boundary probability of the pixels, obviously improves the region conformity and the edge correctness of a segmentation result, and is used for all kinds of image segmentations.

Description

technical field [0001] The invention belongs to the technical field of image processing, and relates to a correction method for segmented images, which can be used for target extraction. Background technique [0002] The active contour model based on curve evolution has been widely used in the field of image segmentation. The active contour model based on curve evolution is a deformable parameter curve controlled by the corresponding energy function, and the deformation of the curve is controlled with the minimization of the energy function as the goal. At present, the combination of the active contour model of curve evolution and the level set method has greatly broadened the application range of the curve active contour model, and has become a widely concerned curve evolution method. This method uses the geometric characteristics of the contour curve to establish the energy functional function of the contour curve movement, minimizes the energy functional function, and ma...

AI Technical Summary

Problems solved by technology

[0004] However, the interactive image segmentation based on the geodesic active contour model has the following disadvantages: for weak sample points in the image such as edges and small areas, the calculated area probability and boundary probability are not accurate enough, resulting in the initial curve eventually evolving into the image The inside of the target, that is, the non-boundary part of the image, resulting in inaccurate image segmentation

Images