A novel statistical golden ratio based adaptive high density impulse noise removal algorithm

Enhancing the quality of images by removing impulse noise is an important preprocessing step while working with images. In this article, a novel and robust algorithm for removal of salt and pepper noise (SAPN) based on statistical golden ratio formula is presented. The proposed method uses a two stage filtering technique to enhance the image quality through removal of SAPN. After detecting the noisy pixels in the image, an adaptive golden ratio based interpolation filtering technique is applied as the Phase-I filter to regenerate the image using the structure and available details of the input image. The image thus obtained in Phase-I is fed to Phase-II where left alpha trimmed average filter is applied to further enhance the image using local features. For performance evaluation, the algorithm is tested on images with corruption levels as high as 99%, and the results obtained is very satisfactory that highlights better performance of the algorithm when compared to the state of the art methods for the entire range of noise density levels.