Key points are not available for this paper at this time.
Recent work in terrain recognition for outdoor mobile robots mainly focused on several typical pure terrain sample classification, and only one terrain feature is extracted for terrain sample description. In this paper, a segmentation scheme for complex terrain samples is designed for the terrain recognition process. The segmentation scheme is achieved using the graph segmentation followed by the watershed segmentation. The terrain image area is identified on the basis of segmentation. For terrain recognition, several global-based features are extracted from the typical terrain samples: Color Histogram, LBP (Local Binary Pattern), CEDD (Color and Edge Directivity Descriptor), with an ELM (Extreme Learning Machine) classifier to implement the classification experiment. We evaluate the performance of the three features and the principle of extracted feature, and we combined color histogram features with LBP features in series method, and the fused feature are proved to be robust to terrain samples with multi-illumination and jitter scenarios with higher classification accuracy.
Liu et al. (Sun,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: