针对三维视频图像的空洞填充中的前景背景分割时容易造成前景对象提取不准确而影响修复效果的问题,提出利用梯度融合与聚类相结合的三维视频图像修复算法.首先利用分水岭算法与标记相结合的办法对图像进行分割;然后充分利用深度图像的...针对三维视频图像的空洞填充中的前景背景分割时容易造成前景对象提取不准确而影响修复效果的问题,提出利用梯度融合与聚类相结合的三维视频图像修复算法.首先利用分水岭算法与标记相结合的办法对图像进行分割;然后充分利用深度图像的深度梯度结构信息,并采用K均值聚类对梯度图像进行标记修正,以增强对前景对象的辨别能力.实验结果表明,该算法较好地克服了原有分水岭算法在图像分割过程中易发生过分割现象,完整地提取了前景对象的纹理信息,使修复图像具有更好的视觉效果,峰值信噪比相比于原算法提高了1~3 d B.展开更多
FUNAABOR-2 is a popular Ofada rice variety grown in a large area under rainfed upland condition across western states of Nigeria. We used the combination of phenotypic and marker-assisted selection(MAS) to improve gra...FUNAABOR-2 is a popular Ofada rice variety grown in a large area under rainfed upland condition across western states of Nigeria. We used the combination of phenotypic and marker-assisted selection(MAS) to improve grain yield of FUNAABOR-2 under drought stress(DS) at the reproductive stage via introgression of two drought quantitative trait loci(QTLs), qDTY12.1 and qDTY2.3. Foreground selection was carried out using peak markers RM511 and RM250, associated with qDTY12.1 and qDTY2.3, respectively, followed by recombinant selection with RM28099 and RM1261 distally flanking qDTY12.1. Furthermore, BC1 F2-derived introgressed lines and their parents were evaluated under DS and non-stress(NS) conditions during the 2015–2016 dry season. Overall reduction of grain yield under DS compared to NS was recorded. Introgressed lines with qDTY12.1 and qDTY2.3 combinations showed higher yield potential compared to lines with single or no QTL under DS, indicating significant positive interactions between the two QTLs under the FUNAABOR-2 genetic background. Pyramiding of qDTY12.1 and qDTY2.3 in the FUNAABOR-2 genetic background led to higher grain yield production under DS and NS.展开更多
基金This work was supported by the National Key Basic Research Program (No. 2006CB102104), the National Natural Science Funda-tion of China (No. 30430500) and the Project of Talent Scientific Research Fund of Henan Science and Technology University (No.05-156).
文摘针对三维视频图像的空洞填充中的前景背景分割时容易造成前景对象提取不准确而影响修复效果的问题,提出利用梯度融合与聚类相结合的三维视频图像修复算法.首先利用分水岭算法与标记相结合的办法对图像进行分割;然后充分利用深度图像的深度梯度结构信息,并采用K均值聚类对梯度图像进行标记修正,以增强对前景对象的辨别能力.实验结果表明,该算法较好地克服了原有分水岭算法在图像分割过程中易发生过分割现象,完整地提取了前景对象的纹理信息,使修复图像具有更好的视觉效果,峰值信噪比相比于原算法提高了1~3 d B.
基金the alliance for a Green Revolution in Africa for providing funds for this study through the West Africa Centre for Crop Improvement, University of Ghana Ph. D fellowship
文摘FUNAABOR-2 is a popular Ofada rice variety grown in a large area under rainfed upland condition across western states of Nigeria. We used the combination of phenotypic and marker-assisted selection(MAS) to improve grain yield of FUNAABOR-2 under drought stress(DS) at the reproductive stage via introgression of two drought quantitative trait loci(QTLs), qDTY12.1 and qDTY2.3. Foreground selection was carried out using peak markers RM511 and RM250, associated with qDTY12.1 and qDTY2.3, respectively, followed by recombinant selection with RM28099 and RM1261 distally flanking qDTY12.1. Furthermore, BC1 F2-derived introgressed lines and their parents were evaluated under DS and non-stress(NS) conditions during the 2015–2016 dry season. Overall reduction of grain yield under DS compared to NS was recorded. Introgressed lines with qDTY12.1 and qDTY2.3 combinations showed higher yield potential compared to lines with single or no QTL under DS, indicating significant positive interactions between the two QTLs under the FUNAABOR-2 genetic background. Pyramiding of qDTY12.1 and qDTY2.3 in the FUNAABOR-2 genetic background led to higher grain yield production under DS and NS.
