Model based methods to compute the fingerprint orientation images were considered more robust and accurate than the gradient based methods as model based methods considered not only the local regions but also the global orientations. This paper presented the stability analysis of the constrained nonlinear phase portrait models of fingerprint orientation images. Thereafter, the constrained nonlinear phase portrait models were revised according to the stability analysis so that they were able to avoid the potential errors due to the stability. Experimental results were presented to demonstrate the revised algorithm.
展开▼