Context-based coding of bilevel images enhanced by digital straight line analysis

Abstract

Abstract-A new efficient compression scheme for bilevel images containing locally straight edges is presented. This paper is especially focused on lossless (intra) coding of binary shapes for image and video objects, but other images with similar characteristics such as line drawings, layers of digital maps, or segmentation maps are also encoded efficiently. The algorithm is not targeted at document images with text, which can be coded efficiently with dictionary-based techniques as in JBIG2. The scheme is based on a local analysis of the digital straightness of the causal part of the object boundary, which is used in the context definition for arithmetic encoding. Tested on individual images of standard TV resolution binary shapes and the binary layers of a digital map, the proposed algorithm outperforms PWC, JBIG, JBIG2, and MPEG-4 CAE. On the binary shapes, the code lengths are reduced by 21%, 27%, 28%, and 41%, respectively. On the map layers, the reductions are 31%, 34%, 32%, and 64%, respectively. The algorithm is also more efficient on the test material than the state-of-the-art generic bilevel image coder free tree.Index Terms-Digital straight line segments, map coding, MPEG-4, object-based video coding, shape coding.

Efficient Coding of Shape and Transparency for Video Objects

“…The DSLSC intra scheme proposed in [2] and [3] is utilized for coding intra pictures in the binary shape layer . The scheme is shortly represented below, followed by the description of the novel extensions for exploiting temporal correlation in subsequent binary shape pictures in inter mode.…”

“…In order to describe the models, a brief introduction to the concept of chain codes and DSLS is required. For details, refer to [3].…”

“…The digital straight line segments coder (DSLSC) [3] was recently proposed for efficient coding of bilevel images with locally straight edges, e.g., single binary shape images and bilevel layers of digital maps. The DSLSC models the edges as digital straight line segments (DSLS) [18], while standard algorithms like JBIG [8], JBIG-2 [9] and MPEG-4 CAE [10], [12] do not fully exploit the information given by the local straightness of the boundary.…”

Coding Transparency in Object-Based Video

“…1. First, the binary shape is directly encoded with DSLSC [3,4]. In order to exploit the correlation between layers, L255 is encoded referencing Lo, and Lint is encoded referencing both Lo and L255, as described in the following sections.…”

“…Shape coding has been studied intensively [1], resulting in the adoption of Context-based Arithmetic Encoding (CAE) into MPEG-4 [2]. The Digital Straight Line Segments Coder (DSLSC) [3,4] was recently proposed as a very efficient alternative to CAE. Although the use of transparency is desirable for smooth composition, little documented work has been aimed at efficient coding of the transparency information [5,6].…”