An algorithm using Hopfield neural network to recognize intersection lines is presented in this paper.This algorithm utilizes a new concept which is much different from the conventional processing methods of intersection lines recognition. In this algorithm

an image containing intersection lines is mapped to a Hopfield neural networks.Every pixel in the image is mapped to a neuron

and the lines between neurons are determined by the relations of the represented pixels such as the distance

potentials difference and slope. During the Hopfield neural nets evaluation

the neurons states are modified till convergence. Along the settlement of the neurons' states

the mapped intersection lines is recognized. During experimental simulations

this algorithm really showed fine performance.