brightness constancy

However, most studies are limited to models with brightness constancy assumptions.

然而，大部分光流研究用的是基于亮度不变假设的模型，而这种假设在不少情况下并不适用。

This interesting paper shows that this illusion "automatically" occurs as a by-product when an artificial neural network is trained for brightness constancy.

这个有趣的文件表明，这种错觉“自动”发生作为副产品时，人工神经网络的亮度恒常训练。

First, we establish the data items of variational optic flow computation, which can effectively calculate optic flow under six cases, such as brightness constancy assumption, brightness gradient constancy assumption and brightness divergence constancy assumption and so on.

首先建立了变分光流计算的数据项，能有效地计算光亮度不变、光亮度梯度不变、光亮度散度不变等假设前提下光流场计算的6种情形，推导了数据项设计的策略和方法。