Abstract:
In past few years, many fields of image processing along with computational photography demand of edge preserving algorithms has increased. Various edge preserving and smoothing algorithms have already been introduced. Image edge preservation algorithm have are mainly splited into two groups: local filters and global optimization filters. Many techniques have been introduced under these categories and they have their own pros and cons. Computational cost of local filters is less but they couldn’t overcome halo near edges where as this problem is overcomed by global filters but many of them create gradient reversal near edges. Local filters work on neighbouring pixels where as global filters work on pixel classes. Still global filters lack in overcoming gradient reversal artifacts which was partially overcomed in Content Adaptive Image Detail Enhancement but few result still aren’t good enough. The existing technique is based on L0 norm to produce a detail enhanced image. This algorithm works on fine details by amplifying them. Amplification is done through enlarging gradients of input image but pixel at edges at not enlarged in order to avoid halos and gradient reversal near edges. Edge aware weighting is also included in L0 norm so that edges are better preserved. There is still need of algorithm that can overcome halos and gradient reversal near edges. A new approach based on existing technique is used in this research which produces better resultant images than the existing technique.
