A New MRCAHE Method for Wear Particle Image Enhancement Based on Improved Online Optical Microfluidic Sensor

Abstract

As direct byproducts of frictional interactions between contacting surfaces, oil wear particles exhibit varied physical properties that provide essential insights into the underlying wear mechanisms and degree of wear severity. Current online optical microfluidic monitoring sensors demonstrate inadequate imaging quality, especially in ferrograph sensors, where particles frequently form chains, hindering real-time monitoring. To address this challenge, an integrated optimization approach has been developed, emphasizing two key aspects: the structural redesign of the online optical sensor and the enhancement of wear particles imaging. We develop a high-precision optical monitoring sensor, which facilitates both conventional particle counting and size detection, as well as the extraction of high-definition texture images of particles in real time. Initially, as the scattering accounts for the majority of the total light energy attenuated, we compute the light intensity scattered by particles in oil. The influence of particle size, scattering angle, particle type, and incident light wavelength on scattering intensity is analyzed, establishing the basis for improving the image quality of wear particles. Then, to alleviate the substantial image degradation induced by oil, characterized by diminished contrast, color attenuation, blurring, and indistinct features, we propose a hybrid image enhancement method MRCAHE, which integrates Multi-scale Retinex with Color Restoration (MSRCR) and Contrast-Limited Adaptive Histogram Equalization (CLAHE). The sensor’s performance is ultimately validated on a high-speed, heavy-load gear fault simulation test bench. Experimental results demonstrate that the sensor consistently collects distinct images of wear particles, and the MRCAHE enhancement method significantly improves deblurring, texture extraction, color restoration, and sharpness. This portable oil wear particle monitoring sensor provides a robust technical foundation for condition monitoring, fault diagnosis, and intelligent maintenance of rotating machinery.