An Interactive Image Segmentation Method

Abstract

The invention discloses an interactive image segmentation method, which is used to solve the problems of low operating efficiency and heavy user workload in the existing interactive image segmentation method. The invention comprises the following steps: the user inputs an image, and selects a rectangular area containing the foreground target image with a rectangular frame; extracts the outer boundary of the rectangular area through the Canny edge detection algorithm; initializes the ternary image with the outer boundary of the foreground target image area, and removes the The background area in the rectangular area, thereby segmenting the foreground target image; output the foreground target image. The invention not only retains the advantage of less user interaction of the GrabCut algorithm, but also makes full use of the boundary information of the foreground object by means of the Canny algorithm, and improves the shortcomings of the GrabCut algorithm in the case of similar foreground / background colors or shadows in the foreground object. In addition, since the method of the invention reduces the number of iterations of the GrabCut algorithm, the operating efficiency is also greatly improved. At the same time, the user is not required to draw a rough outline of the foreground target image, which reduces the workload of the user.

Description

technical field [0001] The invention relates to the technical field of digital image processing, in particular to an interactive image segmentation method. Background technique [0002] Image segmentation is the technology and process of dividing an image into several specific regions with unique properties and proposing objects of interest. The existing image segmentation methods are mainly divided into the following categories: threshold-based segmentation methods, region-based segmentation methods, edge-based segmentation methods, and segmentation methods based on specific theories. [0003] Currently, image segmentation is a key issue in the fields of image processing, pattern recognition, and computer vision. Since the traditional machine automatic segmentation and manual segmentation are difficult to achieve the expected purpose, interactive image segmentation has become the mainstream of current research. Since the development of image segmentation, many interactive...

AI Technical Summary

Problems solved by technology

Among them, the GrabCut method uses the Gaussian mixture model (GMM) to characterize the color probability distribution, and performs image segmentation simply based on the region information. For images with similar color information or shadows in the foreground target, the extracted foreground target is inaccurate and requires a lot of manual correction in the later stage

However, the images of foreground objects that have been artificially corrected in the later stage are often inaccurate, which brings errors to the further analysis and processing of foreground objects.

In response to this problem, Wang Jianqing et al. proposed a target extraction method combining boundaries and regions. This method combines GCBAC and GrabCut algorithms to improve the traditional GrabCut algorithm. Algorithm increases the amount of calculation, and another obvious disadvantage is that this method requires the user to draw the outline of the foreground target, which also increases the workload of the user

Images