Feature Enhancement of Wear Particle Images Based on Curvelet Transform

Compared with wavelet transform, Curvelet transform has characteristics of anisotropy and good curve singularity expression abilities. To advance the validity and reliability of wear particle feature extraction as well as recognition, an image feature enhancement method based on Curvelet transform was proposed. Wear particle images were decomposed into different frequency components by Curvelet transform. Scale enhancement coefficients were introduced into the medium-frequency and high-frequency components to enhance images’ edges and details. Then, enhanced wear particle images were achieved utilizing inverse Curvelet transform. Experiment results indicate that the proposed method can effectively improve image quality. As the details and edges are clear, enhanced images are more suitable for feature extraction and recognition.