A Coarse Representation of Frames Oriented Video Coding By Leveraging Cuboidal Partitioning of Image Data

Abstract

Video coding algorithms attempt to minimize the significant commonality that exists within a video sequence. Each new video coding standard contains tools that can perform this task more efficiently compared to its predecessors. In this work, we form a coarse representation of the current frame by minimizing commonality within that frame while preserving important structural properties of the frame. The building blocks of this coarse representation are rectangular regions called cuboids, which are computationally simple and has a compact description. Then we propose to employ the coarse frame as an additional source for predictive coding of the current frame. Experimental results show an improvement in bit rate savings over a reference codec for HEVC, with minor increase in the codec computational complexity.

Human-Machine Collaborative Video Coding Through Cuboidal Partitioning

“…where j represents different color component channels of the frame C. The feature descriptors m (i) ∈ Z n * are encoded in the way described in [16] and are also part of the bitstream.…”

“…The aforementioned properties of cuboids were leveraged on to traditional video coding in the work of [16,17]; wherein a reference frame is generated using cuboidal approximation of the target frame and incorporating this additional reference frame into a modified HEVC encoder outperformed the ratedistortion performance of a standalone HEVC encoder. The obtained gain in delta PSNR and savings in bit rate can be attributed to the fact that HEVC employs a partitioning scheme that begins at a rigid fixed size of 64 × 64 pixels level and during this process does not take into account the structural properties of the scene need be coded.…”

“…Building on the work [16], in this paper we investigate the applicability of cuboidal features in video coding targeting machines. Based on the CuPID algorithm, (i) features are extracted from a given video frame for the machine vision task of object detection.…”

Dynamic Point Cloud Compression Using A Cuboid Oriented Discrete Cosine Based Motion Model

“…Fig. 1 shows estimated cuboids over the current frame C. In addition to this commonality modeling capability, cuboids are computationally efficient and can be described in a compact way [15].…”

“…In this paper, a commonality modeling paradigm is proposed for the DPC projected texture video sequence, leveraging on cuboids and discrete cosine basis oriented motion model. The contribution of this work are as follows: (i) while existing work [15] estimate the cuboids using current frame's (C) pixel intensity values, we propose to discover them from the difference image (C − R) intensity values where R is the HEVC temporal reference for encoding C. This further improves moving object segmentation in C, (ii) existing approaches [21,22] use the entire frame as support for the dis-crete cosine based motion model which may result in an average motion model. In this work, the support of such motion model is kept limited to individual cuboids only for more accurate motion model, (iii) a cuboid oriented discrete cosine based motion compensated prediction, R cos of the current frame is generated.…”