Loss information recovery of direct unique identification on reflective surface

In order to address the unreadable problem of direct unique identification on metal reflective surface during product real-time tracking, in this paper, two different reflective region images are captured for the same identification, the wavelet-based homomorphic filtering is applied to enhance the contrast of reflective region, the zoning Otsu thresholding is used to get the binary image, the characteristic matrix is extracted and the unique identification information is recovered by feature fusion. The method makes the module misreading rate of product identification fall to 0.7% from 22.91%, avoids the product tracking disconnection, achieves the loss information recovery of direct unique identification on reflective surface, promotes the in-depth application of direct part marking in product life cycle real-time tracking.