Applying Partial Differential Equations on Cubic Uniform Local Binary Pattern to Reveal Micro-Changes

The world we live in everyday, accompany with enormous numbers of minute variations which affect us and our surroundings in several aspects. These variations, so called micro-changes, are low in intensity and brief in duration which makes them almost invisible to naked eyes. Nonetheless, revealing them could open up a new wide range of applications from security, business, engineering, medical, and seismology to psychology. In this paper, we adopted a novel autonomous approach comprising Partial Differential Equations (PDE) and Cubic Uniform Local Binary Pattern (Cubic-ULBP) to spot micro-changes. Cubic-ULBP extracts 15 planes containing robust features against noise and illumination variations. Afterwards, PDE pick out single plane out of 15 to reduce time consumption. The proposed method is not only optimized to get the job done perfectly but also provides promising results comparing with most state-of-the-art methods. So that the accuracy is increased about 36% and 40% on the CASME and the CASME II databases, respectively. The combination of the PDE and the Cubic-ULBP creates a strong and optimal method for detecting the apex frame and micro-movement. This method's performance is found to be promising when the training samples are scarce, too.