Adaptive image enhancement and dynamic-template-matching-based edge extraction method for diamond roller on-machine profile measurement

The complex and harsh environment faced by diamond roller on-machine profile measurement brings challenges to roller image processing technology. Therefore, this paper focuses on the original image enhancement and edge extraction, which has a vital impact on the measurement results of the roller profile in the complex environment. The key research is the multi-scale Retinex (MSR) improved by the fast Deriche filter and the saturation correction method with gating mechanism, and the image enhancement effect with strong edge-preserving ability is realized through parameter optimization. Compared with MSR and several image enhancement methods based on its improvement, the experimental results show that the proposed method obtains a moderate average gray mean, which is 51.0% higher than the original image; it also obtains the highest average standard deviation and average sharpness, which are 21.4% and 58.4% higher than the original image, and 38.5% and 112.9% higher than MSR. Such enhancement effect is most suitable for subsequent roller profile extraction among various comparison methods. Besides, an optimal scale factor is found by optimizing the matching degree to realize the dynamic adjustment of normalized correlation coefficient template. As a result, the interference of wear debris and coolant on the profile are successfully reduced, and the smoother edges of different diamond rollers dressed under various working conditions are obtained. It lays the foundation for realizing high-precision diamond roller on-machine profile measurement.