针对多类不平衡数据分类准确率低的问题,提出一种基于空间扩展的支持向量机学习算法(support vector machine algorithm based on space spreading,SS-SVM)。根据空间扩展原理,在多维欧式空间中通过空间扩展对少类数据进行上采样,使其...针对多类不平衡数据分类准确率低的问题,提出一种基于空间扩展的支持向量机学习算法(support vector machine algorithm based on space spreading,SS-SVM)。根据空间扩展原理,在多维欧式空间中通过空间扩展对少类数据进行上采样,使其处理数据时减少小区块的影响;降低数据不平衡度以优化分类器组;在扩展的数据集上训练SVM分类器。标准数据集上的实验结果表明,与几种经典的算法相比,SS-SVM在多类不平衡数据分类上可获得令人满意的分类结果,对少类数据分类精度要求较高的问题尤为有效。展开更多
目的降采样滤波是生成空间金字塔影像数据的主要手段,但目前没有一种客观指标来鉴别滤波器的降采样效果,因为至少需要空间金字塔的两层原始信号才能计算滤波器的降采样峰值信噪比(PSNR)。为解决此难题,本文建立一种研究路线:先基于视频...目的降采样滤波是生成空间金字塔影像数据的主要手段,但目前没有一种客观指标来鉴别滤波器的降采样效果,因为至少需要空间金字塔的两层原始信号才能计算滤波器的降采样峰值信噪比(PSNR)。为解决此难题,本文建立一种研究路线:先基于视频影像数据评选确定一个性能优秀的降采样滤波器,然后验证该滤波器降采样生成遥感金字塔的主观目视效果,提出一种沿图像纹理方向滤波的降采样方法 TDFA(texture direction filtering approach),可生成高质量的空间影像金字塔。方法本文把降采样与升采样结合提出一种重采样滤波对偶RSFP(re-sampling filter pair),作为当前层金字塔数据的一个逼近,用来评价降采样滤波器效果。基于RSFP评价手段,筛选出一种基于纹理滤波的金字塔生成方法 TDFA:对每个8×8块,TDFA在直流、水平、135°、垂直和45°等5个方向中搜索确定图像的一个纹理方向,用一个3阶滤波器沿纹理方向实施降采样,效果优于目前最好的最邻近插值方法,无任何伪彩、锯齿、块效应或马赛克。结果利用大量影像数据实验,同几个典型滤波器的降采样效果对比,TDFA提升平均PSNR的范围,对拉格朗日滤波器是7.29 8.44 d B;对双线性滤波器是6.26 7.40 d B;对AVS的1/4插值滤波器是5.80 6.84 d B;对最邻近插值是4.51 5.70 d B。结论本文提出的纹理滤波降采样算法可以生成质量优于现有最好水平的遥感金字塔影像,也可以生成高质量的多层视频流媒体数据。所提出的重采样滤波对偶RSFP可以输出当前层的高精度预测,用于可伸缩视频编码处理。展开更多
Scalable video coding (SVC) is a newly emerging standard to be finalized as an extension of H.264/AVC. The most attractive characters in SVC are the inter layer prediction techniques, such as Intra_BL mode. But in c...Scalable video coding (SVC) is a newly emerging standard to be finalized as an extension of H.264/AVC. The most attractive characters in SVC are the inter layer prediction techniques, such as Intra_BL mode. But in current SVC scheme, a uniform up-sampling filter (UUSF) is employed to magnify all components of an image, which will be very inefficient and result in a lot of redundant computational complexity. To overcome this, we propose an efficient component-adaptive up-sampling filter (CAUSF) for inter layer interpolation. In CAUSF, one character of human vision system is considered, and different up- sampling filters are assigned to different components. In particular, the six-tap FIR filter used in UUSF is kept and assigned for luminance component. But for chrominance components, a new four-tap FIR filter is used. Experimental results show that CAUSF maintains the performances of coded bit-rate and PSNR-Y without any noticeable loss, and provides significant reduction in computational complexity.展开更多
文摘针对多类不平衡数据分类准确率低的问题,提出一种基于空间扩展的支持向量机学习算法(support vector machine algorithm based on space spreading,SS-SVM)。根据空间扩展原理,在多维欧式空间中通过空间扩展对少类数据进行上采样,使其处理数据时减少小区块的影响;降低数据不平衡度以优化分类器组;在扩展的数据集上训练SVM分类器。标准数据集上的实验结果表明,与几种经典的算法相比,SS-SVM在多类不平衡数据分类上可获得令人满意的分类结果,对少类数据分类精度要求较高的问题尤为有效。
文摘目的降采样滤波是生成空间金字塔影像数据的主要手段,但目前没有一种客观指标来鉴别滤波器的降采样效果,因为至少需要空间金字塔的两层原始信号才能计算滤波器的降采样峰值信噪比(PSNR)。为解决此难题,本文建立一种研究路线:先基于视频影像数据评选确定一个性能优秀的降采样滤波器,然后验证该滤波器降采样生成遥感金字塔的主观目视效果,提出一种沿图像纹理方向滤波的降采样方法 TDFA(texture direction filtering approach),可生成高质量的空间影像金字塔。方法本文把降采样与升采样结合提出一种重采样滤波对偶RSFP(re-sampling filter pair),作为当前层金字塔数据的一个逼近,用来评价降采样滤波器效果。基于RSFP评价手段,筛选出一种基于纹理滤波的金字塔生成方法 TDFA:对每个8×8块,TDFA在直流、水平、135°、垂直和45°等5个方向中搜索确定图像的一个纹理方向,用一个3阶滤波器沿纹理方向实施降采样,效果优于目前最好的最邻近插值方法,无任何伪彩、锯齿、块效应或马赛克。结果利用大量影像数据实验,同几个典型滤波器的降采样效果对比,TDFA提升平均PSNR的范围,对拉格朗日滤波器是7.29 8.44 d B;对双线性滤波器是6.26 7.40 d B;对AVS的1/4插值滤波器是5.80 6.84 d B;对最邻近插值是4.51 5.70 d B。结论本文提出的纹理滤波降采样算法可以生成质量优于现有最好水平的遥感金字塔影像,也可以生成高质量的多层视频流媒体数据。所提出的重采样滤波对偶RSFP可以输出当前层的高精度预测,用于可伸缩视频编码处理。
基金Supported by China Postdoctoral Science Foundation (Grant No. 20080430454)the Key Laboratory of Geo-informatics of State Bureau of Surveying and Mapping (Grant No. 200834)+1 种基金the National High-Tech Research and Development Program of China (Grant No. 2007AA12Z151)the National Basic Research Program of China (Grant No. 2006CB701303)
文摘Scalable video coding (SVC) is a newly emerging standard to be finalized as an extension of H.264/AVC. The most attractive characters in SVC are the inter layer prediction techniques, such as Intra_BL mode. But in current SVC scheme, a uniform up-sampling filter (UUSF) is employed to magnify all components of an image, which will be very inefficient and result in a lot of redundant computational complexity. To overcome this, we propose an efficient component-adaptive up-sampling filter (CAUSF) for inter layer interpolation. In CAUSF, one character of human vision system is considered, and different up- sampling filters are assigned to different components. In particular, the six-tap FIR filter used in UUSF is kept and assigned for luminance component. But for chrominance components, a new four-tap FIR filter is used. Experimental results show that CAUSF maintains the performances of coded bit-rate and PSNR-Y without any noticeable loss, and provides significant reduction in computational complexity.