Difference between pHash nearest neighbor and color histogram nearest neighbor?

Question

I'm looking to create an animated series of photos that are as similar as possible. While researching, I've come across two methods:

1. Generate a pHash of the images and do a nearest neighbor using the Hamming distance of the hash.
2. Create color histograms and do an n-dimensional nearest neighbor using Euclidian distance.

Many people who have been commenting for the https://stackoverflow.com/questions/6971966/how-to-measure-percentage-similarity-between-two-images question claim that the two processes are essentially the same. I'm looking for a little more insight on this. They seem like different processes.

Thoughts?