Scatter point elimination algorithm suitable for AR virtual soft package synthesis

Abstract

The invention provides a scatter point elimination algorithm suitable for AR virtual soft package synthesis. The scatter point elimination algorithm comprises the following steps: performing image processing operation of firstly corroding and then expanding scattered scatter points in a selected area; positioning the scatter points of which the edges in the selected area are connected into a lineor form scatter point blocks or scatter point zones, calculating the Gaussian weighted sum of the peripheral pixels along the peripheral areas of the scatter points one by one, and adsorbing the scatter points; when commodity replacement operation is carried out, firstly, extracting area selection information to obtain an image M, performing mean filtering processing on the image M, then the imageM serving as a transparency mask, and synthesizing a scene and replaced commodity materials. According to the algorithm, scatter points can be eliminated to the maximum extent. The scatter point elimination rate is 98%, the scatter point misjudgment rate is 10%, stray points are intelligently reduced through a computer, side seams become less obvious after the scatter points are reduced, the influence of the clutter points on vision can be weakened through the eclosion fusion technology in the third step, and meanwhile rendered materials can be better fused into a scene.

Description

technical field [0001] The invention relates to the technical field of AR virtual soft decoration synthesis, in particular to a scatter point elimination algorithm suitable for AR virtual soft decoration synthesis. Background technique [0002] In the AR virtual soft decoration synthesis system, it is necessary to distinguish areas such as walls, floors, windows, etc. The information of these areas needs to be manually calibrated by other tools of the system module. The system uses mask.png to record area information. In the mask, the gray value every 10 The integer represents an area, assuming that the wall area is represented by a gray value of 10-50, the background wall is 60, the floor is 220, the window is 250, etc., such as figure 1 shown; [0003] Different gray values ​​in the selection mask represent different ranges of the selection, but there are a large number of point gray values ​​between the selection and around objects with complex edges such as bonsai that ...

AI Technical Summary

Problems solved by technology

The number of scattered points varies depending on the number of constituencies and the complexity of the constituencies. figure 2 The number of scatter points has reached more than 1900. Too many scatter points will cause very thin black / white edges to appear on the edges after the change, such as image 3 As shown, it affects the visual experience, although the small black / white borders are not obvious, but when it is projected on a high-definition screen or a large screen, such small defects will stand out and directly affect the visual experience. Therefore, the present invention proposes a A scatter elimination algorithm suitable for AR virtual soft decoration synthesis to solve the deficiencies in the prior art

Images