英语翻译We also try other values ranging from 2 to 12 for a between feature1 and 3 at a step of 0.1 for every validation image.We find thatthe best results are achieved with a around 3 and 5,and the quantitativedifference is trivial at a 0.001 sc
来源:学生作业帮助网 编辑:六六作业网 时间:2024/12/30 03:16:33
英语翻译We also try other values ranging from 2 to 12 for a between feature1 and 3 at a step of 0.1 for every validation image.We find thatthe best results are achieved with a around 3 and 5,and the quantitativedifference is trivial at a 0.001 sc
英语翻译
We also try other values ranging from 2 to 12 for a between feature
1 and 3 at a step of 0.1 for every validation image.We find that
the best results are achieved with a around 3 and 5,and the quantitative
difference is trivial at a 0.001 scale.What’s more,the
P2PME of feature 1 and feature 3 in Fig.7 can be viewed as the results
when a becomes extremely small and extremely large,
respectively.These two extremes yield obviously worse results
than the reasonable range of alpha.So our hypothesis for a also
holds in practice.
With the same way as (Wei and Yeung,2007),we also draw a
histogram to show correlation coefficients Rðxl;xhÞ between the
reconstructed weight vectors xl for low-resolution patches and
weight vectors xh for high-resolution patches.It can be seen from
Fig.4 that most of the correlation coefficients are located in [0.5,1],
which shows the features we select from the patch are effective
and reasonable
英语翻译We also try other values ranging from 2 to 12 for a between feature1 and 3 at a step of 0.1 for every validation image.We find thatthe best results are achieved with a around 3 and 5,and the quantitativedifference is trivial at a 0.001 sc
We also try other values ranging from 2 to 12 for a between feature 1 and 3 at a step of 0.1 for every validation image.
我们还为每个确认的图象以0.1的步进、针对特征1到3 之间,尝试从2到12的其他值范围修正.
We find that the best results are achieved with a around 3 and 5,and the quantitative difference is trivial at a 0.001 scale.
我们发现在3到5周围达到最佳结果,且数量差异细微到0.001范围.
What’s more,the P2PME of feature 1 and feature 3 in Fig.7 can be viewed as the results when a becomes extremely small and extremely large,respectively.
还有,图示7中的特征1和3的P2PME可以分别显示当变化趋于极小和极大时的结果.
These two extremes yield obviously worse results than the reasonable range of alpha.So our hypothesis for a also holds in practice.
这两个极端结果明显较差于alpha的合理范围.因此我们的假设也只是个对实践的把握.
With the same way as (Wei and Yeung,2007),we also draw a histogram to show correlation coefficients Rðxl;xhÞ between the reconstructed weight vectors xl for low-resolution patches and weight vectors xh for high-resolution patches.
采用同样的方法(魏和Yeung,2007年),我们也绘制了个柱状图,显示在低分辨图象碎片的重向量xl和高分辨图象碎片的重向量xh之间重构的相关性系数Rðxl;xhÞ.
It can be seen from Fig.4 that most of the correlation coefficients are located in [0.5,1],which shows the features we select from the patch are effective and reasonable
从图示4中可以看到,大部分相关性系数位于[0.5,1],这表示我们从碎片中所选的特征是有效而合理的.