Toshio Kamei

Image filter design for fingerprint enhancement

Fingerprint filter design and a method of enhancing fingerprint image are discussed. Two distinct filters in the Fourier domain are designed, a frequency filter corresponding to ridge frequencies and a direction filter corresponding to ridge directions on the basis of fingerprint ridge properties. An energy function for selecting image features, i.e. frequencies and directions, is defined by intensities of images obtained with the above filters and a measure of smoothness of the features. Using the image features which minimize the energy function, an enhanced image is produced from the filtered images. This image enhancement method is applied to fingerprint matching. In experiments with rolled prints, it is shown that the matching error rate is reduced by about two-thirds compared with Asais method.

Face retrieval by an adaptive Mahalanobis distance using a confidence factor

This paper proposes an adaptive Mahalanobis distance for face retrieval. The distance is derived from a posterior distribution of observation errors in features categorized by confidence of face images. Since the distance is calculated considering error variances of each dimension according to the confidence, it can reflect error distribution of each matching more precisely than a standard Mahalanobis distance. We apply this distance to eigenface techniques using image contrast and asymmetric components of face images as the confidence. To evaluate our proposed distance in face retrieval, we made experiments using MPEG-7 face descriptors as eigenface features. The best match ratio was improved from 93.5% to 97.6% compared with the weighted distance described in MPEG-7 by using the proposed distance.

Fingerprint card classification with statistical feature integration

This paper describes a fingerprint classification algorithm for an automated fingerprint identification system with a large-size ten-print card database. The classification algorithm determines a fingerprints pattern category based on a ridge structure analysis and a direction-based neural network, and computes additional feature characteristics such as core-delta distance, along with confidence indexes associated with each feature. A card preselector then integrates the set of obtained features after weighting them according to the features expected error and inherent selection capability, calculates the card similarity based on feature differences, statistically evaluates the conditional probability of each pair being a correct match, and selects the most similar subset of the database as candidates for minutiae matching. The experimental results confirm that effective classification capability of 0.2% false acceptance with 2% false rejection has been achieved.

Spectral eigenfeatures for effective DP matching in fingerprint recognition

Dynamic Programming (DP) matching has been applied to solve distortion in spectral-based fingerprint recognition. However, spectral data is redundant, and its size is huge. PCA could be used to reduce the data size, but leads to loss of topographical information in projected vectors. This allows only inter-vector similarity estimations such as Euclid or Mahalanobis distances, and proves to be inadequate in presence of distortion occurring in finger sweeping with a line sensor. In this paper, we propose a novel two-step PCA to extract compact eigen-features amenable to DP matching. The first PCA extracts eigenfeatures of Fourier spectra from each image line. The second extracts eigen features from all lines to form the feature templates. In matching, the feature templates are inversely transformed to line-by-line representations on the first PCA subspace for DP matching. Fingerprint matching experiments demonstrate the effectiveness of our proposed approach in template size reduction and accuracy improvement.