Target Recognition of Ladar Range Images Using Slice Image: Comparison of Four Improved Algorithms
Wenze Xia,Shaokun Han,Jingya Cao,Liang Wang,Yu Zhai,Yang Cheng
DOI: https://doi.org/10.1117/1.oe.56.7.073107
IF: 1.3
2017-01-01
Optical Engineering
Abstract:Compared with traditional 3-D shape data, ladar range images possess properties of strong noise, shape degeneracy, and sparsity, which make feature extraction and representation difficult. The slice image is an effective feature descriptor to resolve this problem. We propose four improved algorithms on target recognition of ladar range images using slice image. In order to improve resolution invariance of the slice image, mean value detection instead of maximum value detection is applied in these four improved algorithms. In order to improve rotation invariance of the slice image, three new improved feature descriptors-which are feature slice image, slice-Zernike moments, and slice-Fourier moments-are applied to the last three improved algorithms, respectively. Backpropagation neural networks are used as feature classifiers in the last two improved algorithms. The performance of these four improved recognition systems is analyzed comprehensively in the aspects of the three invariances, recognition rate, and execution time. The final experiment results show that the improvements for these four algorithms reach the desired effect, the three invariances of feature descriptors are not directly related to the final recognition performance of recognition systems, and these four improved recognition systems have different performances under different conditions. (C) 2017 Society of Photo-Optical Instrumentation Engineers (SPIE)