Research Of 3-D Reconstruction Of Fiame Based On Computer Stereo Vision

3-D reconstruction of industrial flame could be very useful in industrial field.3-D reconstruction of industrial flame can help people understand the fullness of flame and flame temperature distribution which is helpful to study the flame propagation mechanism and the development of the model of dynamic simulation research. As the flame flicks and has no regular evident characteristic,3-D reconstruction of flame is difficult and lack of study. Some reconstruction ways by tomographic imaging, matching the dense point cloud texture have been used, but these ways are limited by the cost and time.Aiming at these deficiencies and practical application,3-D reconstruction of industrial flame is studied to fill the defect of the less.The research has certain economic and leading effect.In this paper, the main research contents are as follows:introduces stereo vision system and the structure and principle in detail. This system using eight image acquisition from different perspectives for real-time shooting a candle flame to obtain its gray images.Then, according to the proposed flame three-dimensional reconstruction algorithm,3-D reconstruction of industrial flame can be rebuild.According to the characteristics of foreground and background in industrial flame grey image which has show the continuous block distribution, a trend aggregation function is proposed to detect the different areas where the intensity of the image is changed rapidly.The flame feature points are selected from these areas. The experimental results shows the validity and feasibility of Region Feature Point Detection Algorithm in Industrial Flame Based on Tropism (FPDAT). Veracity and time-taken of this proposed method is 5.5% higher and 13.4% shorter than traditional algorithms.In this paper, one stereo matching algorithm for flame based on distance ratio constraint (DRCFMA) is proposed which has better performance on matching time and accuracy. The algorithm takes flame characteristics into account and uses some operators to get the ratio constraint of matched feature points.The algorithm is applied on the following 3-D reconstruction of flame. The validity and feasibility of DRCFMA are shown by the experimental results.This paper put two innovation alogorithms into 3D reconstruction.The final 3D experiment shows that the result has clear and continuously changing greyscale layers.To sum up, this paper proposes a fast and accurate 3D reconstruction method of flame. Experiments prove that this method is the supplement of the existing fire 3D reconstruction methods.