The current draft scalable extension to HEVC offers twoapproaches, RefIdx and TextureRL, for performing inter-layer prediction.In the framework of TextureRL, this paper first presents a mode-dependentpixel-based weighted intra prediction scheme for coding the enhancementlayer (EL). The scheme first decomposes the EL intra prediction and the collocated base layer reconstructed block into their respective DC andAC components and then computes a weighted sum of both to form a better prediction signal using a pixel-based weighting scheme. Theweighting factors to associate with different components are obtained by a least-quares fit to the training data. It was observed that they dependstrongly on the EL?s intra prediction mode and prediction block size,but are
less dependent on QP settings. The second step is to introduce a simplified mode with respect to the analysis of the coding
parameters to reduce the storage requirement. The experimental results show an average BD-rate savings of 1.0% for theAI-2x
configuration and 0.5% for AI-1.5x over the SHM-1.0 anchor.