A method based on local feature point detecting and matching for force measurement of micro-cantilever was proposed. Images of micro-cantilever before and after deformation were obtained by optical microscope. Based on the enlarged speckle texture of the cantilever surface, the LOG (Laplace of Gaussian) feature points with the local extreme response value were detected in the scale space. An affine-invariant closed region was extracted for each LOG feature point. Then the centroid of this closed region could be calculated and treated as a feature point with subpixel positioning. Affine-invariant feature descriptors could be constructed and used for the matching of the corresponding points in images before and after deformation. The deformation of the microcantilever could be described by deflection curve obtained by fitting the displacement of the matching points， and then the force value as well as position applied by the loader on the cantilever were calculated using the least square method. The performance of the proposed method was verified by experiments based on actual deformation of microcantilever.