Fusion parametric image enhanced micro‐damage edge gradient extraction in ceramic bearings

Abstract To address the issue of insufficient accuracy in identifying gradual edge features of micro‐damage in ceramic bearings, this paper proposes an image enhancement method that integrates fast Fourier transform (FFT) and parameter self‐correction guided filtering. The approach involves developing an FFT‐based frequency‐domain separation model to effectively extract high‐frequency detailed textures from the image for enhanced edge characterization. Meanwhile, by incorporating the local variance properties of noise distribution, a self‐correcting parameter‐guided filtering model is established to strengthen target boundaries and suppress noise interference. Experimental results demonstrate that the processed images achieve a peak signal‐to‐noise ratio of 44.5971 dB and a signal‐to‐noise ratio of 89.4046, significantly improving the discernibility and extraction completeness of micro‐damage edge features. This study provides an effective image processing solution for the non‐destructive testing of micro‐damage in ceramic bearings.