Unsupervised segmentation of color-texture regions in images and video

Abstract

ÐA new method for unsupervised segmentation of color-texture regions in images and video is presented. This method, which we refer to as JSEG, consists of two independent steps: color quantization and spatial segmentation. In the first step, colors in the image are quantized to several representative classes that can be used to differentiate regions in the image. The image pixels are then replaced by their corresponding color class labels, thus forming a class-map of the image. The focus of this work is on spatial segmentation, where a criterion for ªgoodº segmentation using the class-map is proposed. Applying the criterion to local windows in the class-map results in the ªJ-image,º in which high and low values correspond to possible boundaries and interiors of color-texture regions. A region growing method is then used to segment the image based on the multiscale J-images. A similar approach is applied to video sequences. An additional region tracking scheme is embedded into the region growing process to achieve consistent segmentation and tracking results, even for scenes with nonrigid object motion. Experiments show the robustness of the JSEG algorithm on real images and video.

Video Segmentation for Markerless Motion Capture in Unconstrained Environments

“…Chapter Four presents an in-depth description of the proposed segmentation framework. It highlights the major procedures involved in the 7 segmentation and tracking as well as describes how the technique has been modified and enhanced from the original algorithm proposed by Deng et al [3]. Chapter Five discusses the experimental setup of the framework and its performance.…”

“…This JSEG clustering procedure proposed by Deng et al [3] introduced a serious limitation for the application of the algorithm on real life images. This limitation was identified by Wang et al in [28].…”

“…As mentioned before, the technique is based on Deng and Manjunath's JSEG implementation [3] with key improvements in order to make it more appropriate to the context considered here. Our algorithm is described as a set of five key processes:…”

“…Deng et al [3] have proposed a segmentation technique which not only relies on colour information but also on texture data in order to cut a scene into semantic regions.…”

“…Another interesting entry into these types of techniques consists of the modifications that Deng et al [3] added to their JSEG algorithm in order to allow it to segment multiple frames in a batch process. While initially the technique does not lend itself very well to video block segmentation, its authors have introduced a supplementary temporal segmentation criterion named the //-value.…”