Improved BM3D by using region growing segmentation & Sobel edge technique

• Published in: 2014 7th International Congress on Image and Signal Processing pp. 164 - 168
• Main Authors: Rehan Afzal, H. M., Ali, Husan, Asif, Muhammad
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-10-2014
• Subjects: Arrays; Image Denoising; Image edge detection; Image restoration; Image segmentation; Improved BM3D; Noise; Noise measurement; Noise Removal; Remove blurriness; Satellites; Three-dimensional displays
• DOI: 10.1109/CISP.2014.7003770

In image processing to achieve de-noising is an important task. The main thing in image de-noising is to attain edges of image. Many algorithms and methods have been proposed recently but they have not yet achieved the desired level. Among these block matching 3D (BM3D) is the method which is considered best to remove noise as compared to previous techniques. Before BM3D technique, Non Local Means technique was introduced which is a very simple technique with normal results. This work is the expansion of BM3D de-noising technique, in which region growing segmentation technique has been used. Also Sobel edge operator is added to improve the results. The proposed algorithm shows very good results for standard as well as satellite images in terms of PSNR and removal of blurriness, it is very good for sharpness of image; edges of satellite images should be very clear and the proposed algorithm achieves this feature.