A New Hybrid Edge Adaptive Algorithm For Blocking Artifact Reduction On High Definition Video

Hamid, Mohd Saad and Darsono, Abd Majid and Hamzah, Rostam Affendi and Kadmin, Ahmad Fauzan and Abd Ghani, Shamsul Fakhar (2018) A New Hybrid Edge Adaptive Algorithm For Blocking Artifact Reduction On High Definition Video. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (24 Pages) A New Hybrid Edge Adaptive Algorithm For Blocking Artifact Reduction On High Definition Video.pdf - Submitted Version Restricted to Registered users only Download (503kB)

Abstract

The main objective for this research is to perform study on characteristic of edge adaptive filter algorithm for edge-preserving smoothing of noisy images.Fundamentally, the Sum of Absolute Differences (SAD) algorithm produces an accurate result on the stereo video processing for the textured regions.However,this algorithm sensitive to low texture and radiometric distortions (i.e., contrast or brightness).To overcome these problems,the proposed algorithm utilizes edge-preserving filter which is known as Bilateral Filter (BF).The BF algorithm reduces noise and sharpen the images.Additionally,BF works fine on the low or plain texture areas.The hybrid model for filter which combine the features from edge adaptive filter and BF will be designed.Then produced video quality also will be measured qualitatively and quantitatively in a computing environment by using stereo image sample.This research will help to provide better quality of reconstructed image which will help to improve the output image.The hybrid implementation of the algorithm focused on cost aggregation and disparity map refinement stage in stereo matching algorithm.The experiments are carried out on the platform of Window 10 on desktop PC with 3.2GHz processor and 8GB memory.To evaluate the accuracy,the experimental images are using a standard quantitative online stereo benchmarking dataset from the Middlebury and qualitative measurement based on the stereo video of autonomous navigation benchmarking dataset from KITTI. Based on the results,the proposed algorithm is among the lowest of average errors which indicates the competitive achievement of the proposed work.Furthermore,the proposed framework is competitive with some established algorithms and the proposed work is also well capable to work with real environment of vehicle navigation which was displayed in the published paper.

Actions (login required)

View Item