Spray characterisation using combined radon and hough transforms

The paper presents an improved injector spray characterization method using image analysis. The existing methods of image analysis such as granulometry and Hough transform are not efficient. The size of heavily overlapped spray particles cannot be accurately determined using granulometric methods. The Hough transform method is better in separating the overlapped droplets. But it produces erroneous data with smaller droplet sizes. Here, an improved method using both Radon and Hough transforms is presented to detect drop size distribution. In this method, Radon transform is used to differentiate between the overlapped and non-overlapped droplets and Hough transform to separate overlapped droplets. The proper segmentation of droplet regions are important in the accurate computation of Sauter mean diameter (SMD). As the segmentation by thresholding is affected by noise, an active contour based image segmentation using distance regularized level set evolution (DRLSE) is used for the segmentation of droplet regions. The proposed method is used to compute drop size distribution from the image of spray and corresponding histogram is also plotted. The result obtained is illustrated to be better than that obtained using Hough transform alone. Methods to detect spray cone angle and breakup length are also illustrated in this paper.