针对单一特征步态识别率低的问题,提出一种将步态能量图(Gait Energy Image,GEI)中动态部分和Gabor小波特征融合的步态识别算法.首先,通过运动目标检测及二值化和形态学处理等预处理操作得到步态轮廓图,再进一步从步态轮廓图计算得到步...针对单一特征步态识别率低的问题,提出一种将步态能量图(Gait Energy Image,GEI)中动态部分和Gabor小波特征融合的步态识别算法.首先,通过运动目标检测及二值化和形态学处理等预处理操作得到步态轮廓图,再进一步从步态轮廓图计算得到步态能量图,并从中分割出动态部分.然后,利用Gabor小波从步态能量图的动态部分中提取不同角度的信息,将两步态特征融合在一起,对融合后得到的特征向量用改进的KPCA方法进行降维.最后,将降维后的融合特征向量输入到基于多分类的支持向量机(Support Vector Machine,SVM)中,从而完成步态的分类和识别.经过在中国科学院自动化研究所CASIA步态数据库上进行实验,取得了很好的识别效果,实验结果表明,与单一特征的步态识别方法相比,融合后算法的识别率提高了约10%.展开更多
Accurate forecasting of wind velocity can improve the economic dispatch and safe operation of the power system. Support vector machine (SVM) has been proved to be an efficient approach for forecasting. According to th...Accurate forecasting of wind velocity can improve the economic dispatch and safe operation of the power system. Support vector machine (SVM) has been proved to be an efficient approach for forecasting. According to the analysis with support vector machine method, the drawback of determining the parameters only by experts' experience should be improved. After a detailed description of the methodology of SVM and simulated annealing, an improved algorithm was proposed for the automatic optimization of parameters using SVM method. An example has proved that the proposed method can efficiently select the parameters of the SVM method. And by optimizing the parameters, the forecasting accuracy of the max wind velocity increases by 34.45%, which indicates that the new SASVM model improves the forecasting accuracy.展开更多
为了预防驾驶员因疲劳驾驶引发交通事故,提出一种基于改进的HOG(histogram of oriented gra⁃dients)算法与CNN-SVM结合的人脸疲劳检测方法。首先,针对传统HOG特征提取过程中的不足,对HOG特征提取时加入了对角线像素的灰度信息,使得提取...为了预防驾驶员因疲劳驾驶引发交通事故,提出一种基于改进的HOG(histogram of oriented gra⁃dients)算法与CNN-SVM结合的人脸疲劳检测方法。首先,针对传统HOG特征提取过程中的不足,对HOG特征提取时加入了对角线像素的灰度信息,使得提取到的灰度边缘信息更加丰富,从而提升对人脸检测的准确度。其次,使用级联回归(ensemble of regression trees,ERT)算法对脸部68个关键点标定并裁剪眼部区域。最后,使用CNN结合SVM对人眼状态识别,并结合PERCLOS等参数进行疲劳判断。展开更多
文摘针对单一特征步态识别率低的问题,提出一种将步态能量图(Gait Energy Image,GEI)中动态部分和Gabor小波特征融合的步态识别算法.首先,通过运动目标检测及二值化和形态学处理等预处理操作得到步态轮廓图,再进一步从步态轮廓图计算得到步态能量图,并从中分割出动态部分.然后,利用Gabor小波从步态能量图的动态部分中提取不同角度的信息,将两步态特征融合在一起,对融合后得到的特征向量用改进的KPCA方法进行降维.最后,将降维后的融合特征向量输入到基于多分类的支持向量机(Support Vector Machine,SVM)中,从而完成步态的分类和识别.经过在中国科学院自动化研究所CASIA步态数据库上进行实验,取得了很好的识别效果,实验结果表明,与单一特征的步态识别方法相比,融合后算法的识别率提高了约10%.
基金Project(71071052) supported by the National Natural Science Foundation of ChinaProject(JB2011097) supported by the Fundamental Research Funds for the Central Universities of China
文摘Accurate forecasting of wind velocity can improve the economic dispatch and safe operation of the power system. Support vector machine (SVM) has been proved to be an efficient approach for forecasting. According to the analysis with support vector machine method, the drawback of determining the parameters only by experts' experience should be improved. After a detailed description of the methodology of SVM and simulated annealing, an improved algorithm was proposed for the automatic optimization of parameters using SVM method. An example has proved that the proposed method can efficiently select the parameters of the SVM method. And by optimizing the parameters, the forecasting accuracy of the max wind velocity increases by 34.45%, which indicates that the new SASVM model improves the forecasting accuracy.
文摘为了预防驾驶员因疲劳驾驶引发交通事故,提出一种基于改进的HOG(histogram of oriented gra⁃dients)算法与CNN-SVM结合的人脸疲劳检测方法。首先,针对传统HOG特征提取过程中的不足,对HOG特征提取时加入了对角线像素的灰度信息,使得提取到的灰度边缘信息更加丰富,从而提升对人脸检测的准确度。其次,使用级联回归(ensemble of regression trees,ERT)算法对脸部68个关键点标定并裁剪眼部区域。最后,使用CNN结合SVM对人眼状态识别,并结合PERCLOS等参数进行疲劳判断。
