Neural Network Segmentation of Laser Scanning System Data
DOI:
https://doi.org/10.22213/2413-1172-2017-3-125-129Keywords:
neural network, clouds of points, segmentation, laser scanning systemAbstract
Objects detection and tracking is one of the perspective and rapidly developing areas of technical vision, applied in various systems. Clusterization and segmentation of data are an integral part of such systems. The paper presents a brief overview of the 2D laser scanning systems data segmentation algorithms. The segmentation of closely located objects is one of the weaknesses of such algorithms. Application of the range data along with the intensity values of the corresponding reflected signal measurement was suggested for better segmentation. For the data fusion, it was proposed to use a neural network, as a convenient tool for clustering and segmentation of noisy and complex data. In this paper a multilayer perceptron is used as one of the most common and studied models. The architecture of the received network, the method of its training and the interpretation of the output for segmentation of the cloud of two-dimensional points are presented. The results of this study can be used not only in the problems of segmentation of clouds of points, but also in solving the problems of mapping, for the feature point detection in the analysis of two-dimensional clouds of points, for developing the systems of objects tracking.References
Video segmentation by tracking many figure-ground segments / F. Li [et al.] // Proceedings of the IEEE International Conference on Computer Vision. - 2013. - P. 2192-2199.
Ochs P., Malik J., Brox T. Segmentation of moving objects by long term video analysis // IEEE transactions on pattern analysis and machine intelligence. - 2014. - Vol. 36, No. 6. - P. 1187-1200.
Fully automated segmentation and tracking of the intima media thickness in ultrasound video sequences of the common carotid artery / D. E. Ilea [et al.] // IEEE transactions on ultrasonics, ferroelectrics, and frequency control. - 2013. - Vol. 60, No. 1. - P. 158-177.
Yao W., Hinz S., Stilla U. Object extraction based on 3d-segmentation of lidar data by combining mean shift with normalized cuts : Two examples from urban areas // Urban Remote Sensing Event. - 2009 Joint. - P. 1-6.
Rubio D. O., Lenskiy A., Ryu J. H. Connected Components for a Fast and Robust 2D Lidar Data Segmentation // Modelling Symposium (AMS). - 2013, 7th Asia. - IEEE, 2013. - P. 160-165.
Efficient segmentation of 3d lidar point clouds handling partial occlusion / J. Aue, D. Langer, B. Muller-Bessler, B. Huhnke // Intelligent Vehicles Symposium (IV), 2011 IEEE, 2011. - P. 423-428.
Difference of normals as a multi-scale operator in unorganized point clouds / Y. Ioannou, B. Taati, R. Harrap, M. Greenspan. - ArXiv e-prints, Sep. 2012.
Simultaneous segmentation and superquadrics fitting in laser-range data / R. Pascoal [et al.] // IEEE Transactions on Vehicular Technology. - 2015. - Vol. 64, No. 2. - P. 441-452.
Thuy M., Leon F. P. Non-linear, shape independent object tracking based on 2d lidar data // Intelligent Vehicles Symposium, 2009 IEEE. - IEEE, 2009. - P. 532-537.
Rubio D. O., Lenskiy A., Ryu J. H. Op. cit.
Fast reciprocal nearest neighbors clustering / R. J. López-Sastre [et al.] // Signal Processing. - 2012. - Vol. 92, No. 1. - P. 270-275.
Schuster H. F. Segmentation of lidar data using the tensor voting framework // International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences. - 2004. - Vol. 35, No. B3. - P. 1073-1078.
Simultaneous segmentation and superquadrics fitting in laser-range data / R. Pascoal [et al.] // IEEE Transactions on Vehicular Technology. - 2015. - Vol. 64, No. 2. - P. 441-452.
Large-scale video classification with convolutional neural networks / A. Karpathy [et al.] // Proceedings of the IEEE conference on Computer Vision and Pattern Recognition. - 2014. - P. 1725-1732.
Chen Y., Zhao X., Jia X. Spectral-spatial classification of hyperspectral data based on deep belief network // IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. - 2015. - Vol. 8, No. 6. - P. 2381-2392.
