Chapter

Pattern Recognition

Volume 8142 of the series Lecture Notes in Computer Science pp 435-445

Efficient Multi-cue Scene Segmentation

  • Timo ScharwächterAffiliated withEnvironment Perception, Daimler R&D
  • , Markus EnzweilerAffiliated withEnvironment Perception, Daimler R&D
  • , Uwe FrankeAffiliated withEnvironment Perception, Daimler R&D
  • , Stefan RothAffiliated withDepartment of Computer Science, TU Darmstadt

* Final gross prices may vary according to local VAT.

Get Access

Abstract

This paper presents a novel multi-cue framework for scene segmentation, involving a combination of appearance (grayscale images) and depth cues (dense stereo vision). An efficient 3D environment model is utilized to create a small set of meaningful free-form region hypotheses for object location and extent. Those regions are subsequently categorized into several object classes using an extended multi-cue bag-of-features pipeline. For that, we augment grayscale bag-of-features by bag-of-depth-features operating on dense disparity maps, as well as height pooling to incorporate a 3D geometric ordering into our region descriptor.

In experiments on a large real-world stereo vision data set, we obtain state-of-the-art segmentation results at significantly reduced computational costs. Our dataset is made public for benchmarking purposes.