The Study Of Fast Intra Algorithms For Spatial Scalable High-efficiency Video Coding

In recent years,video coding technology has developed rapidly.However,with the diversity of terminal equipment,network heterogeneity and differences of user demand,the existing video coding technology cannot meet many requirements.Therefore,scalable extension of HEVC,known as Scalable High-efficiency Video Coding(SHVC)has been developed.The coding efficiency of SHVC is very high,but the complexity of coding procedure is also very high,which causes a great challenge for practical applications.In order to improve the coding speed with negligible loss in coding efficiency,three different fast encoding strategies aiming to spatial scalable video coding are proposed.The main work and innovation points are as follows:1.Depths of the current encoding unit are predicted based on the spatial and interlayer correlations.According to the distribution of the base layer,videos are divided into simple,complex and medium sequences.A weight value is set for each adjacent coding unit,According to the degree of correlation between the current encoding unit and adjacent coding units.Weight values are combined with depths of the relevant coding unit to predict likelihood depth and eliminate depths with small possibilities.2.Early terminations are proposed based on distribution of residual coefficients.Based on the distributions of residual coefficients,the conditions of early terminations are calculated to improve the coding speed.3.Directional modes are predicted and early terminated.The likelihood directional modes are predicted based on directional modes of neighboring coding units.Based on weights of the direction modes,an early termination of directional modes are proposed to further improve coding speed.Experimental results show that the algorithms can significantly improve the coding speed with negligible loss of coding efficiency.