期刊文献+
共找到3篇文章
< 1 >
每页显示 20 50 100
基于FPGA的低复杂度快速SIFT特征提取
1
作者 姜晓明 刘强 《北京航空航天大学学报》 EI CAS CSCD 北大核心 2019年第4期804-810,共7页
尺度不变特征变换(SIFT)算法具有优良的鲁棒性,在计算机视觉领域得到广泛应用。针对SIFT算法高计算复杂度而导致其在CPU上运行实时性低的问题,基于现场可编程门阵列(FPGA)设计了一种低复杂度的快速SIFT硬件架构,主要对算法的特征描述符... 尺度不变特征变换(SIFT)算法具有优良的鲁棒性,在计算机视觉领域得到广泛应用。针对SIFT算法高计算复杂度而导致其在CPU上运行实时性低的问题,基于现场可编程门阵列(FPGA)设计了一种低复杂度的快速SIFT硬件架构,主要对算法的特征描述符提取部分进行优化。通过降低梯度信息(包括梯度幅值和梯度方向)的位宽、优化高斯权重系数的产生、简化三线性插值系数的计算和简化梯度幅值直方图索引的求解等方法,避免了指数、三角函数和乘法等复杂计算,降低了硬件设计复杂度和硬件资源消耗。实验结果显示,提出的低复杂度快速SIFT硬件架构,与软件相比,可以获得约200倍的加速;与相关研究相比,速度提高了3倍,特征描述符稳定性提高了18%以上。 展开更多
关键词 现场可编程门阵列(FPGA) 尺度不变特征变换(SIFT) 硬件设计 梯度信息 特征描述符提取
下载PDF
Novel histogram descriptor for global feature extraction and description 被引量:3
2
作者 张刚 马宗民 +1 位作者 邓立国 徐长明 《Journal of Central South University》 SCIE EI CAS 2010年第3期580-586,共7页
A novel histogram descriptor for global feature extraction and description was presented. Three elementary primitives for a 2×2 pixel grid were defined. The complex primitives were computed by matrix transforms. ... A novel histogram descriptor for global feature extraction and description was presented. Three elementary primitives for a 2×2 pixel grid were defined. The complex primitives were computed by matrix transforms. These primitives and equivalence class were used for an image to compute the feature image that consisted of three elementary primitives. Histogram was used for the transformed image to extract and describe the features. Furthermore, comparisons were made among the novel histogram descriptor, the gray histogram and the edge histogram with regard to feature vector dimension and retrieval performance. The experimental results show that the novel histogram can not only reduce the effect of noise and illumination change, but also compute the feature vector of lower dimension. Furthermore, the system using the novel histogram has better retrieval performance. 展开更多
关键词 feature extraction and description histogram descriptor gray histogram edge histogram
下载PDF
Modified Fourier descriptor for shape feature extraction 被引量:1
3
作者 张刚 马宗民 +1 位作者 牛连强 张纯明 《Journal of Central South University》 SCIE EI CAS 2012年第2期488-495,共8页
A modified Fourier descriptor was presented. Information from a local space can be used more efficiently. After the boundary pixel set of an object was computed, centroid distance approach was used to compute shape si... A modified Fourier descriptor was presented. Information from a local space can be used more efficiently. After the boundary pixel set of an object was computed, centroid distance approach was used to compute shape signature in the local space. A pair of shape signature and boundary pixel gray was used as a point in a feature space. Then, Fourier transform was used for composition of point information in the feature space so that the shape features could be computed. It is proved theoretically that the shape features from modified Fourier descriptors are invariant to translation, rotation, scaling, and change of start point. It is also testified by measuring the retrieval performance of the systems that the shape features from modified Fourier oescriptors are more discriminative than those from other Fourier descriptors. 展开更多
关键词 shape feature extraction Fourier descriptors centroid distance approach
下载PDF
上一页 1 下一页 到第
使用帮助 返回顶部