Modified hybrid median filter for removal of low density random-valued impulse noise in images / Dr. Siti Noraini Sulaiman ... [et al.]

Digital images are apt to deteriorate during acquisition, storage and transmission in noise environment. Impulse Noise (IN) is one of the common noises which are prevalent in noisy communication channels, faulty pixel sensors and memory location. Its affect the accuracy of various image processing applications such as image segmentation, image restoration, image classification and image compression. There are two common types of IN models defined in image processing; Fixed-Valued Impulse Noise (FIN) and Random-Valued Impulse Noise (RVIN). The FIN model is easier to restore compared to RVIN model as the RVIN normally occurs randomly across the image. Median filter is the most popular choice for removing the IN due to its efficiency and high computational efficiency. However, the performance drops at high noise densities, resulting the appearance of patches and unpreserved image. Another major drawbacks is, it normally been implemented uniformly across the whole image without being consider whether the treated pixels are "noisy-pixels" or "noise-free pixels". Therefore, this project is intended to address the conflicting issue through formulating new RVIN removal algorithm based on two phase iterative method; adaptive-switching filtering methodology and local preserving scheme. The switching phase act as detector to identify, locate and distinguish "noisy-pixels" by utilizing the histogram information of the corrupted images and fuzzy reasoning concept. The incorporation of fuzzy is to handle uncertainty present in the corrupted images. The detected "noisy-pixels" will then be subjected to the next stage of preserving action, while the "noise-free pixels" are reserved and left untreated. The preserving stage will utilize local information based on belongingness concept in order to provide an accurate outcome; the noise kept attenuated and the original image is preserved. The fundamental findings from this research can be applied in many applications such as in medical imaging (radiology department (MRI image)) and telecommunication (satellite image).