A novel computation method for 2D deformation of fish scale based on SURF and N-R optimisation

Abstract

Fish scales were unique structural materials that served as a form of natural armour and affected the mechanical properties which had important applications in bionics. Digital image correlation (DIC) method was used to determine the mechanical properties, but it took a long time to calculate the uniaxial tensile deformation. In this investigation a DIC optimisation algorithm method based on speeded-up robust features (SURF) and Newton-Raphson (N-R) was conducted on specimens prepared from the scales. First, the SURF algorithm was used to detect the matched points and their coordinate values in the digital images before and after deformation. Then, the initial displacement of the interest point was estimated from the affine transformation fitted to the matched feature points inside the subset area. Last, the zero-mean normalised sum of squared differences (ZNSSD) metric function was optimised by the N-R iterative method, and the optimised displacement value of the interest points would be gained. The numerical translation experiments and simulation results showed that this method improved the search speed and the measurement accuracy effectively. So the deformation of fish scales for axial tension would be calculated by this method.