Optimum Color and Contrast Enhancement for Online Ferrography Image Restoration

Abstract

Online ferrography, because of its nondestructive and real-time capability, has been increasingly applied in monitoring machine wear states. However, online ferrography images are usually degraded as a result of undesirable image acquisition conditions, which eventually lead to inaccurate identifications. A restoration method focusing on color correction and contrast enhancement is developed to provide high-quality images for subsequent processing. Based on the formation of a degraded image, a model describing the degradation is constructed. Then, cost functions consisting of colorfulness, contrast, and information loss are formulated. An optimal restored image is obtained by minimizing the cost functions, in which parameters are properly determined using the Lagrange multiplier. Experiments are carried out on a collection of online ferrography images, and results show that the proposed method can effectively improve the image both qualitatively and quantitatively.