Research And Application Of Fuel Spray Image Matching

Usually in the study of fuel spray image, we should do some pretreatment on the fuel spray image in Photoshop software to raise the image definition. After that,statistics characteristics of particles and record the data experimental analysis needed artificially.The paper adopt direct shot method to real-time shoot the spray of No.O diesel oil, with the manual fuel injection pump, high resolution digital camera, nozzle of vehicle engine et., After getting the image, artificially statistics and record the characteristic data particles. Finally, using the artificially statistical data calculation the particle various features and drawing the graph experimental needed.The paper proposed a way that using image matching in the studying of fuel spray image. By integrating the spray image characteristics, Comparative analysis of several kinds of traditional image matching algorithms.The traditional SSDA matching algorithm is sensitive to gray, its ability is poor when resistance to the noise and rotation; but its computing speed is fast.The NCC algorithm remedy the shortage of the SSDA algorithm. But the matching calculation is too big and the speed is too slow. The paper proposed a new way of image matching which besides on the sequential similarity detection algorithm (SSDA). The paper integrate two algorithms into one, then using wavelet decomposition to realize multi-resolution image matching. Finally, using the pixel sampling methods to improve the search way of image matching. New algorithm not only keeping the matching accuracy but also improve the matching speed. And then use the MATLAB/GUI software to programming the new algorithm, calculation the statistical characteristics of particle and drawing variety curve the experiment needed. Finally, test the software performance by comparing the results using the software and manual.