High-precision simple interactive image matting method, storage device and terminal

Abstract

The invention discloses a high-precision simple interactive image matting method, a storage device and a terminal. The method includes the steps: inputting an original image to be matted; outputting prompt messages according to different use environments; acquiring user operation commands and determining operation modes; performing calculation to obtain a tripartite graph; acquiring foreground and background sample point sets of unknown region pixels; estimating non-transparency of the unknown region pixels and taking non-transparency with the highest confidence coefficient as image matting non-transparency when extracting sample points; performing smoothing operation on neighborhoods of the unknown region pixels to obtain image matting non-transparency of the smoothed unknown region pixels; extracting a foreground target from the original image according to the smoothed image matting non-transparency and color values of sample point pairs. The method is simple in operation and applicable to the field of image processing, and has high image matting precision.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a high-precision and simple interactive map-cutting method, a storage device and a terminal. Background technique [0002] Matting refers to the process of precisely separating the foreground area of ​​interest from the image or video background. It is widely used in video editing, virtual reality, film production and other fields. Through simple fusion processing after matting, the foreground object can be combined with various backgrounds at will, greatly reducing the workload of scene layout. [0003] According to the theory of natural image matting, each pixel in the image can be represented by formula (9): [0004] I=αF+(1-α)B (9) [0005] Among them: I represents the color value in the actual image, F represents the foreground color value, B represents the background color value, and α is called the foreground opacity. [0006] The input image to be cut can be divided int...

AI Technical Summary

Problems solved by technology

The traditional blue-screen matting is simple and easy, but requires the target to be placed in a blue background, which limits the shooting scene and scope of use; Knockout matting is suitable for areas with smooth transitions in color, but local detail matting is poor; Bayesian matting And Possion matting are suitable for situations where the foreground and background are quite different, but when the foreground color distribution is complex, the matting effect is poor; Robust matting is more stable, but it is often powerless for non-connected edges

In addition, the above algorithms generally have the defects of high computational complexity and difficulty in real-time application

[0008] Existing image-matching software such as Photoshop are relatively complicated to operate, require certain professional knowledge, and are time-consuming and labor-intensive; foregrounds extracted by TouchRetouch and Map-Making Master often have insufficient fine edges and more burrs

Images