针对带钢表面缺陷识别率受到光照变化、纹理复杂多样以及噪声干扰而导致误识别率高的问题,提出一种新的带钢表面缺陷识别算法。首先从增加邻域联系的角度改进多块局部二值模式(MB-LBP)特征,缓解提取过程中因所选子窗口尺寸大小不同而造...针对带钢表面缺陷识别率受到光照变化、纹理复杂多样以及噪声干扰而导致误识别率高的问题,提出一种新的带钢表面缺陷识别算法。首先从增加邻域联系的角度改进多块局部二值模式(MB-LBP)特征,缓解提取过程中因所选子窗口尺寸大小不同而造成的保留图像细节与去除噪声之间的平衡性问题;其次将改进的MB-LBP特征与梯度方向直方图(HOG)特征线性加权得到融合特征,弥补MB-LBP特征没有表征缺陷边缘和方向的缺点,从而更全面地表征复杂的缺陷纹理;最后通过同时增加全局信息和监督信息改善的局部保持投影(LPP)算法将高维的融合特征非线性映射到低维的本质特征空间中,减少融合特征冗余对分类器识别率的影响。在NEU数据集上仿真实验结果表明:算法对光照变化、纹理复杂多样、以及噪声具有一定的鲁棒性,在信噪比为50 d B情况下将带钢表面缺陷识别准确率提高了5. 17%。展开更多
A new distributed node localization algorithm named mobile beacons-improved particle filter （MB-IPF） was proposed. In the algorithm, the mobile nodes equipped with globe position system （GPS） move around in the wi...A new distributed node localization algorithm named mobile beacons-improved particle filter （MB-IPF） was proposed. In the algorithm, the mobile nodes equipped with globe position system （GPS） move around in the wireless sensor network （WSN） field based on the Gauss-Markov mobility model, and periodically broadcast the beacon messages Each unknown node estimates its location in a fully distributed mode based on the received mobile beacons. The localization algorithm is based on the IPF and several refinements, including the proposed weighted centroid algorithm, the residual resampling algorithm, and the markov chain monte carlo （MCMC） method etc., which were also introduced for performance improvement. The simulation results show that our proposed algorithm is efficient for most applications.展开更多
文摘针对带钢表面缺陷识别率受到光照变化、纹理复杂多样以及噪声干扰而导致误识别率高的问题,提出一种新的带钢表面缺陷识别算法。首先从增加邻域联系的角度改进多块局部二值模式(MB-LBP)特征,缓解提取过程中因所选子窗口尺寸大小不同而造成的保留图像细节与去除噪声之间的平衡性问题;其次将改进的MB-LBP特征与梯度方向直方图(HOG)特征线性加权得到融合特征,弥补MB-LBP特征没有表征缺陷边缘和方向的缺点,从而更全面地表征复杂的缺陷纹理;最后通过同时增加全局信息和监督信息改善的局部保持投影(LPP)算法将高维的融合特征非线性映射到低维的本质特征空间中,减少融合特征冗余对分类器识别率的影响。在NEU数据集上仿真实验结果表明:算法对光照变化、纹理复杂多样、以及噪声具有一定的鲁棒性,在信噪比为50 d B情况下将带钢表面缺陷识别准确率提高了5. 17%。
文摘A new distributed node localization algorithm named mobile beacons-improved particle filter （MB-IPF） was proposed. In the algorithm, the mobile nodes equipped with globe position system （GPS） move around in the wireless sensor network （WSN） field based on the Gauss-Markov mobility model, and periodically broadcast the beacon messages Each unknown node estimates its location in a fully distributed mode based on the received mobile beacons. The localization algorithm is based on the IPF and several refinements, including the proposed weighted centroid algorithm, the residual resampling algorithm, and the markov chain monte carlo （MCMC） method etc., which were also introduced for performance improvement. The simulation results show that our proposed algorithm is efficient for most applications.
