Model based object recognition for particle size distribution

Abstract

In Aerospace applications, Ammonium Nitrate, Ammonium dichromate [(NH4)2 Cr2 O7] and Hydroxy-Terminator PolyButadiene (HTPB) are used as solid propellants. Ammonium dichromate must be pulverized and used in rockets which play the major role to improve the burn rate of the fuel. Image Processing based techniques are used to measure the particle size of Ammonium dichromate. If the particle is overlapped its size or shape can't be accurately measured. Hence, overlap of the particle is determined and its size can be measured. This paper deals with the estimation of the area of Ammonium dichromate particles. An intimate approach to overcome overlap in the particles is done using the Gaussian process regression model. The prediction is done by using linear function of the observed values and these functions can be computed quickly using recursive formulas. Gaussian process regression model is employed to determine the multiple overlap of the particles.