My thoughts about Optical flow has stayed at the late 80s for a long time. For me, optical flow equals to Lukas-Kanade. Until very recent, I found that there are many other good optical flow methods published in top conferences, like CVPR, ICCV. There is also a library – middlebury to test some tough problems within the computer vision area, such as stereo, optical flow. Moreover, the optical flow code written by Prof. Michael Black is very reliable, http://www.cs.brown.edu/~black/.