Recently, the 3D reconstruction technology is widely used to model objects, analyze surface and structure of objects. One of the most popular research topics is the reconstruction for flame, it is more intuitive to research temperature field and burning performance of the flame. However, it is difficult to reconstruct for the flame due to its property of flickering and no-significance. Thus, it is important to develop an effective way to reconstruct it.Follows shows the outline of this paper:Firstly, the significance of 3D reconstruction for flame is analyzed, and so much research results obtained in the world are introduced. Meanwhile, we also compare some traditional methods of reconstruction for the flame.Secondly, an improved reconstruction method for flame is developed based on SIFT algorithm. We add a characteristic weight for flame when selecting feature points, it can help to search feature points more fast and accurately, which overcomes the drawbacks of poor application of traditional feature points selection methods in the flame.Thirdly, a novel reconstruction method for flame is developed based on the thermal radiation inversion of flame algorithm. It compares gray histogram and radiation characteristic curve, and combines Genetic Algorithm with thermal radiation inversion, the optimal solution can be obtained, thus, a good 3D model of flame can be reconstructed.Finally, a flame image collection platform with multi-CCD splitting is designed, which use the synchronous filtered images to reconstruct vertical flame field through thermal radiation inversion. Experiments are performed to verify feasibility of our method, from which we can see that the reconstruction flame models have a clear hierarchy and with continuous gray. |