Research And Implementation On Complex Scene Modeling And Optimization Technology Of Educational Robot

Virtual reality technology in the past few years has been payed more attention byscholars at home and abroad, and it has broad application in the future. Thethree-dimensional scene modeling is an important part of virtual reality technology.Scene modeling techniques, modeling optimization techniques is the two mainlyresearch direction of three-dimensional scene modeling. Scene modeling technology hasa building modeling method, texture mapping, light and material, and other relatedtechnologies, modeling optimization technology has a scene segmentation technology,visibility culling, level of detail, and other related technologies.This article is intended to achieve the educational robot platform scene subsystem,complex game scenarios used to build virtual robots to be used in simulation athletics.This article research, achieve and improved several common scenario modelingtechniques and modeling optimization techniques, the following major has thesecontent:1.We research three modeling methods and select the geometric modeling methodto build an object model and export it into obj file by3Ds MAX modeling software forscenes system; research wood grain texture, the wave texture synthesis formula oftexture material of the the scenes system in the timber, lakes, mip-map technology tosolve the anti-aliasing problem caused by the stretching and contraction of texture;achieve the phong simple illumination model, considering the specular reflection,diffuse reflection and ambient light reflection.2. Improve the BSP tree scene segmentation algorithm, we structure the originalmodel before the division, after the division we refine the structured model, so that weuse these two models to complete the division of the scene, to overcome the problemthat in the original devision operation each side of the object space can not existconcave polygon, to reduce the excess depth and devision operation of the scene tree;we also research the technology of instancing, when we create many similar objects inthe scene, we use the abstract model object pattern to store model information, usingthese abstract objects to enrich have their own size, position, direction, color, in this way we save the storage space of the model.3.Using a hierarchical occlusion graph to improve the extended shadow clippingalgorithm, we establish hierarchical occlusion graph of multi-resolution to be applied tothe two tests of overlap and depth, this algorithm saves a lot of time on the test of beingcovered.4. Improve the multi-resolution model storage of the traditional LOD technology,we only retain the storage information of the highest resolution model, and save thedifference between each layer of the resolution model to set different logos on thehighest resolution model of storage information, comparing to the original storagemodel the stored all of the resolution model the new method saves lots of the storagespace, and we found that the algorithm should be smoother when switching betweendifferent resolution of model.