The shape of Huanghua pear is one of the most important features in classification. Images of Huanghua pears were acquired by means of machine vision system． The method to describe the shape of irregular fruit was studied, in which the Fourier transform and Fourier inverse transform were applied． A sort of software for fruit shape identification based on artificial neural network was developed. It was concluded that the first sixteen harmonics of the Fourier descriptor were enough to represent the primary shape of pear, and the shape identification accuracy could reach 90 % by applying the Fourier descriptor in combination with artificial neural network. Furthermore, this method can also be used to identify the shape of other fruits as the suitable training set is found.