The Alternative Abnormality Discovery and Recovery Algorithm Rooted on MDBUT-Mean Filter for CAIA

For abnormality discovery and recovery of data in research and industry community, especially digital image data, the abnormality discovery and recovery algorithms are the necessary process for digital image processing. The philosophy document proposes the alternative abnormality discovery and recovery algorithm rooted on the Modified decision based unsymmetric trimmed mean filter (MDBUT-Mean Filter) that is upgraded from mean filter in order to overcome the deficiency of the preliminary abnormality discovery and recovery algorithms, for consistent amplitude impulsive abnormality (CAIA). Particular abnormality pixel, which is philosophically resolved from the its abnormality pixel and their nearby pixels by MDBUT-Mean Filter, is discovered to be whether a normality pixel or an abnormality pixel by integrating with the pre-defined calculation area of all nearby pixels. In the investigated simulation, profuse abnormality illustrations, which are collected of Girl and Baboon at all range of abnormality consistency, are implemented for investigating the skillfulness of MDBUT-Mean Filter for abnormality discovery and recovery algorithm. The first contribution of this paper is the simulated determination of the optimal pre-defined calculation area of all nearby pixels in order to make the MDBUT-Mean Filter the highest discovery rate: abnormality discovery rate, normality discovery rate, miss rate and fault alarm. The second contribution of this paper is the contrastive skillfulness investigation of the MDBUT-Mean Filter, which is contrasted with the preliminary abnormality recovery algorithms: SMF, SF and AMF in objective indicator (PSNR) and subjective indicator (illustration quality).