Automatic Motion And Obstacle Avoidance Of Camera For 3D Mobile Animation

The three-dimensional(3D) Animation Auto-generation Technology of Mobile phone aims to deal with monotonous SMS issues with references to Academician R.Q Lu's Full Life Cycle Computer Aided Animation Generation as well as the popular 3G and 4G techs. The system based on this technology manages to convert the sender's message into a 3D animation before reaching the receiver's cell phone one by means of a series of artificial intelligence technologies.The 3D scene camera planning plays a basic part in this system, since the camera planning generates a direct impact on the 3D animation, all the images of which are rendered through the scene cameras. In this system, the images a good camera records should be coherent and vivid movements, able to capture the scene model that can best manifest the text messages, and to avoid obstacles that will affect the visual effects in the scene.There are three versions of camera planning of this system. In view of the former two versions, in this paper, the author make use of the camera primitives to do camera motion planning, camera obstacle detection and obstacle avoidance, and finally reach a judgment objectively based on targets' visibility. The main work is as follows:1) Camera motion planning built upon the camera primitives. For the first time, the author designs the camera motion planning with the camera primitives in the system. The author designs and implements the camera primitive, divides the motion planning into targets selection and motion. The author classifies the models added to the scene in the plot planning, and gets target groups in terms of the forms and expressiveness of their messages and summarizessets of camera movements. Afterwards, varied photographic skills are in correspondence with the target groups while the sets of camera movements are described in details. The camera motion planning can be qualitatively depicted since the basic camera primitives are eventually acquired. In the quantitative part, the author works out a conversion mode to translate the qualitative description to quantitative data, and takes steps to calculate the quantitative data to achive the camera planning.2) Camera obstacle avoidance and targets visibility analysis. The author marks 3D information formodels in the knowledge base. As for the camera qualitative planning of every short message, the scenario models are divided into targets and obstacles. Then for the first time, the author takes the solid geometry as well as the actual situation facing the system into consideration to design and carry out the camera obstacle detection method and mitigation strategies adopts the approach of bounding box for decrease on the detection complexity, reduces the three-dimensional interferences into the two-dimensional intersection judgment, and ajust the camera when occlusion occurs. A method is made to judge the target visibility after an analysis calculation of the Maya camera parameters. In this method, one is supposed to respectively calculate whether the target is within the scope of the camera in the light of the camera position and the visual center in the 3D space coordinates, and then adjust the camera by comparing the obtained statistics with the defined stantards.Analyze the experimental data one year after the new version's updating. First of all, overally analyze the the 1212 text messages. Then the three experiments, namely the diversity experiment of camera planning, the obstacle avoiding and targets visibility experiments and the opening experiments, are conducted based on the research objective. The rate of different results of camera planning is 88%, and for single frame, the rate of no obstacle is 95%, obstacle avoidance success rate is 80%, enough targets can be seen at a rate of 86% and the rate of adjust invisible targets successfully is 55.56%. In the opening experiment, other participants evaluate the auto-selected test data in a specified scope, the satisfaction degree of the animation reaches 66.73%, and the acceptance of camera obstacle reaches 69.94%.The author puts forward a camera motion planning which is based on camera primitives, designs and implements camera obstacle detection and avoidance algorithm, and creates a target visibility judgment method in this system. The findings show that the camera planning has made greater improvement in camera motion and obstacle avoidance, and the visibility judgment promotes the connection between the qualitative planning and the quantitative calculation. The author does be aware that there is still a lot more to do in the future, enlarging the knowledgebase, combining the camera planning with the machine learning together with statistical methods to foster the animation automatical system.