| 摘要 |
An automatic fingerprint enrollment and identification technique is provided for automatically processing an input image represented by pixels having either a one or a zero value. The automatic processing involves a series of steps in which the initial processing is a rough screening technique for eliminating a reasonable portion of the noise while passing through almost all of the minutia. As the steps proceed, the processing becomes more complex to provide a tighter screening out of noise while passing through as much of the minutia as possibe. The later steps by operating on relatively fewer minutia or anomaly points that the earlier steps can be economical though more complex so as to be more discriminate than the earlier steps. The processing provides an optimum trade off between enrolling as large a proportion of the population as possible while minimizing the number of false identification errors. |
