关键词: 图像去噪, 图像识别, 非下采样Shearlet变换, 概率阈值去噪

Abstract: In digital image processing，the requirements for the input image are high，and if there exists noise and interference in the input image，the feature extraction and the following detection and recognition will be inaccurate. To solve the problem，this paper proposed an image denoising algorithm based on non-subsampled shearlet transform. The algorithm used non-subsampled shearlet transform to decompose the image from multiple dimensions and directions，which could describe the detailed information better，such as outline and curve of the image. Besides，the algorithm utilized the coefficients after the threshold value decomposition，to achieve noise removal. The results showed that the algorithm could keep the detailed information of image after noise removal，which contributed to the image detection and recognition.

Key words: image denoising, image recognition, non-subsampled shearlet transform, probability threshold denoising