Object three-dimensional modeling and rendering system as well as generation and rendering methods of three-dimensional model

Abstract

The invention discloses an object three-dimensional modeling and rendering system as well as generation and rendering methods of a three-dimensional model. The system comprises a stepping motor rotation platform, a motor control system, an image collection and processing system, a background plate apparatus and an image collection apparatus. The generation method of a three-dimensional model comprises the following steps: step one, carrying out system initialization; step two, collecting a 360 degree video under a first background plate; step three, collecting a 360 degree video under a second background plate; and step four, utilizing a dual-mask background difference method to carry out cutting off of backgrounds frame by frame. In addition, the rendering method comprises the following steps that: a stored video model is decoded; according to mouse control of a current user, a visual angle that the current user expects is resolved; if a difference between a current visual angle and a visual angle of a certain frame in the video is less than a threshold e, the frame of the video is decoded and the decoded frame is used as a current rendering image; if the difference between the current visual angle and the visual angle of the certain frame in the video is not less than the threshold e, two frames of images that are closest to the user expected angle are selected and linear image interpolation is carried out, so that a rendering image of the user expected angle is obtained.

Description

technical field [0001] The invention relates to an object three-dimensional modeling and rendering system based on IBR (Image Based Rendering, image-based rendering) technology and a three-dimensional model generation and rendering method, belonging to the fields of virtual reality, computer graphics and pattern recognition. Background technique [0002] A core problem in virtual reality technology is to generate high-fidelity images of complex virtual reality environments in real time. In traditional virtual reality methods, to generate a virtual geometric image of an object, first use the geometric graphics rendering method (Graph-based Rendering, GBR) to perform geometric modeling and paste textures, such as modeling using 3DMax and Maya The tool carries out the modeling of 3D objects; then uses the principle of perspective projection in the 3D engine to transform the 3D model into a 2D image pixelated and displayed in the 2D screen space. [0003] However, for complex o...

AI Technical Summary

Problems solved by technology

[0014] 1. The method of image acquisition is still manual, and the image is captured by art editing, so the efficiency is low, and it is not convenient for high-density sampling, shooting and processing. Basically, more than a dozen pictures are used to describe an object;

[0015] 2. Background subtraction relies on manual processing, which is basically processed by software such as PS, without automatic processing;

[0016] 3. When the viewpoint selected by the user is not within the range of the reference image, there is no strategy for synthesizing new viewpoints, and the image required by the user’s current viewpoint cannot be generated. The rendering method is relatively passive, and only the captured images can be browsed.

[0026] 1. It is highly dependent on lighting conditions, requires uniform and constant lighting, and generally has poor adaptability;

[0027] 2. When the color of the foreground object is consistent with the background color, some parts of the object (parts close to the background color) may be subtracted as the background, resulting in a "hollow out" phenomenon;

[0028] Typical "hollow out" phenomenon, such as attached figure 2 As shown, because the color of a certain part of the foreground object is very close to the background color, the generated mask image itself has "hollowout", and the final image also appears "hollowout" when it is synthesized. When this happens, manual work is required in the later stage. Repair the "hollow out" part, resulting in low efficiency and poor adaptability

The inherent defects of the traditional background difference algorithm have led to the lack of an automated rapid modeling and rendering IBR system in the current application field

Images