By employing the Hirota’s bilinear method and different test functions, the breather solutions of HSI equation with different structures are obtained based on symbolic calculation with perturbation parameters. Some n...By employing the Hirota’s bilinear method and different test functions, the breather solutions of HSI equation with different structures are obtained based on symbolic calculation with perturbation parameters. Some new lump solitons are found in the process of studying the degradation behavior of breather solutions. The interaction between lump solution and soliton solution is constructed in the form of lump solution, and the motion trajectory of lump is obtained. In addition, the theorem of lump solitons and N-solitons superposition is given and proved. The superposition formula of lump is derived from the theorem, and its spatial evolution behavior is given.展开更多
针对自然条件下光照条件变化给大田油菜图像分割带来的问题,该文研究了油菜图像的高斯HI颜色分割算法,为作物生长发育周期的自动识别提供前期准备。已有统计结果表明,在仅保留绿色作物的图像中,不同色调值的像素数量服从高斯分布。该文...针对自然条件下光照条件变化给大田油菜图像分割带来的问题,该文研究了油菜图像的高斯HI颜色分割算法,为作物生长发育周期的自动识别提供前期准备。已有统计结果表明,在仅保留绿色作物的图像中,不同色调值的像素数量服从高斯分布。该文将去掉背景信息的样本数据从RGB颜色模型转换至HSI颜色模型后,统计各个光强的所有像素对应的色调值,并计算其期望值和方差,依次得出所有强度所对应色调值的期望值和方差,建立出油菜作物色调强度查找表(hue intensity-look up table)。在此基础上,计算每个像素的色调值和期望值之间的差值,若差值小于阈值,则像素被分割为作物,否则为背景。为了在高斯HI颜色分割算法中确定合适的阈值,该研究选取了45幅不同天气状况(晴天、阴天和雨天)不同发育阶段(苗期、三叶期和四叶期)的油菜图像作为样本,探讨阈值的选取与分割结果的关系。结果表明阈值在[2.4,2.6]内分割效果最佳,油菜目标的形状特征完整度最好。为了对图像分割结果进行评价,分别利用高斯HI颜色模型、CIVE(color index of vegetation extraction)、EXG-EXR(excess green-excess red)、EXG(excess green)和VEG(vegetation)算法对15幅不同天气状况的图像进行分割。从视觉效果上来看,高斯HI算法仅需少量样本,即可达到满意分割效果。与其他方法相比,高斯HI颜色分割算法的误分割率(misclassification error,ME)仅为1.8%,相对目标面积误差(relative object area error,RAE)仅为3.6%,均优于其他4种算法的试验结果。在分割结果稳定性上,高斯HI颜色算法表现最好,其ME和RAE值的标准差最低,分别为0.7%和4.5%。试验结果表明,高斯HI颜色算法能取得较好的分割效果,而且对光照条件变化并不敏感,同时,能够充分保留油菜形状特征的完整性,为后期油菜生长发育周期的自动识别提供可靠数据。展开更多
文摘By employing the Hirota’s bilinear method and different test functions, the breather solutions of HSI equation with different structures are obtained based on symbolic calculation with perturbation parameters. Some new lump solitons are found in the process of studying the degradation behavior of breather solutions. The interaction between lump solution and soliton solution is constructed in the form of lump solution, and the motion trajectory of lump is obtained. In addition, the theorem of lump solitons and N-solitons superposition is given and proved. The superposition formula of lump is derived from the theorem, and its spatial evolution behavior is given.
文摘针对自然条件下光照条件变化给大田油菜图像分割带来的问题,该文研究了油菜图像的高斯HI颜色分割算法,为作物生长发育周期的自动识别提供前期准备。已有统计结果表明,在仅保留绿色作物的图像中,不同色调值的像素数量服从高斯分布。该文将去掉背景信息的样本数据从RGB颜色模型转换至HSI颜色模型后,统计各个光强的所有像素对应的色调值,并计算其期望值和方差,依次得出所有强度所对应色调值的期望值和方差,建立出油菜作物色调强度查找表(hue intensity-look up table)。在此基础上,计算每个像素的色调值和期望值之间的差值,若差值小于阈值,则像素被分割为作物,否则为背景。为了在高斯HI颜色分割算法中确定合适的阈值,该研究选取了45幅不同天气状况(晴天、阴天和雨天)不同发育阶段(苗期、三叶期和四叶期)的油菜图像作为样本,探讨阈值的选取与分割结果的关系。结果表明阈值在[2.4,2.6]内分割效果最佳,油菜目标的形状特征完整度最好。为了对图像分割结果进行评价,分别利用高斯HI颜色模型、CIVE(color index of vegetation extraction)、EXG-EXR(excess green-excess red)、EXG(excess green)和VEG(vegetation)算法对15幅不同天气状况的图像进行分割。从视觉效果上来看,高斯HI算法仅需少量样本,即可达到满意分割效果。与其他方法相比,高斯HI颜色分割算法的误分割率(misclassification error,ME)仅为1.8%,相对目标面积误差(relative object area error,RAE)仅为3.6%,均优于其他4种算法的试验结果。在分割结果稳定性上,高斯HI颜色算法表现最好,其ME和RAE值的标准差最低,分别为0.7%和4.5%。试验结果表明,高斯HI颜色算法能取得较好的分割效果,而且对光照条件变化并不敏感,同时,能够充分保留油菜形状特征的完整性,为后期油菜生长发育周期的自动识别提供可靠数据。