The invention discloses a background suppression method and system based on neighborhood structures. The realization of the method includes: obtaining structure vectors on the basis of gray-scale values of neighborhood pixel points, calculating similarity measures of the structure vectors and neighborhood structure templates, dividing the neighborhood structure templates into a target template and background templates, and utilizing the similarity measures to obtain a background likelihood value of being a background point by an pixel point and a target likelihood value of being a target point by the pixel point; determining the pixel point as the target point if the ratio of the target likelihood value to the background likelihood value is greater than or equal to a threshold value, and otherwise, determining the pixel point as the background point; and using a gray-scale mean value of the background points in a neighborhood pixel point set as an estimated background value, carrying out direct suppression if the pixel point is the background point, and subtracting the estimated background value from a gray-scale value of the target point for background suppression if the pixel point is the target point. The method has a good suppression effect and self-adaptability for different background types, and satisfies real-time calculation requirements at the same time.