The invention relates to a method for de-noising of a balanced morphology filter image optimized by a particle swarm, comprising the following steps of: firstly, defining an unit structure element of a zero square matrix, a balanced erosion operation and a balanced expansion operation; secondly, updating speeds and positions of particles by a particle swarm optimization technology after a peak signal-to-noise ratio is taken as a cost function, so as to enable a transformed value of the global optimum particle position to be as the size of the unit structure element; and finally, performing the balanced erosion operation on an image by the optimized unit structure element, then performing the balanced expansion operation, and outputting the image after the balanced expansion operation. The method for de-noising of the balanced morphology filter image optimized by the particle swarm, disclosed by the invention, overcomes the inherent shortages in an extreme value operation used by the present morphology, can adaptively obtain the size of the unit structure element according to the degree of pollution on the image caused by noise and effectively remove pulse noise in the image, as well as can be widely used in the fields of image processing, licence plate information extracting, edge detecting etc.