Medical Image Denoising Using Sub Band Adaptive Thresholding Techniques Based on Wavelet 2D Transform

Medical images are corrupted by noises during its transmission and acquisition process. Noise reduction has been a traditional problem in image and signal processing. Medical images generally contains minute information about heart, brain, nerves etc therefore wrong diagnosis might not rescue the patient from harmful effects. In this paper we proposed an approach for image denoising based on wavelet 2D transform using adaptive thresholding technique. The proposed technique estimates the threshold value and decomposition level for an image. In this an additive white Gaussian noise is added to image and forward wavelet transform is applied on noisy image. After this wavelet coefficients are threshold and inverse wavelet transformation is performed to restore the original image. The proposed method reduce the noise from image more effectively. The MATLAB result shows that adaptive thresholding method is better than the other traditional methods as it minimize the mean square error (MSE). Bayes soft thresholding obtained better results in terms of PSNR value.