Curvelet transform based image denoising via Gaussian mixture model

This paper presents a novel image denoising method based on curvelet transform and gaussian mixture model. After decomposing noisy images into curvelet domain, gaussian mixture model (GMM) is applied and obtained statistical parameters are used for calculating adaptive level depended thresholds. Noise removal is performed using hard threshold method in the curvelet coefficients of each sub-band. Due to the adaptive thresholding for each level the restored images are visually satisfactory.