Binary Two-Dimensional PCA

Yanwei Pang, Dacheng Tao, Yuan Yuan, Xuelong Li

Abstract: Fast training and testing procedures are crucial in biometrics recognition research. Conventional algorithms, e.g., <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">principal</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">component</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">analysis</i> (PCA), fail to efficiently work on large-scale and high-resolution image data sets. By incorporating merits from both <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">two-dimensional</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">PCA</i> (2DPCA)-based image decomposition and fast numerical calculations based on Haarlike bases, this technical correspondence first proposes <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">binary</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2DPCA</i> (B-2DPCA). Empirical studies demonstrated the advantages of B-2DPCA compared with 2DPCA and binary PCA.