I'm sure you can find detailed definition over internet. Here is my attempt to simplify things.
Recursion is calling itself again.
Dynamic Programming is a way to solve problems which exhibit a specific structure (optimal sub structure) where a problem can be broken down into sub problems which are similar to original problem. Clearly one can invoke recursion to solve a DP. But it is not necessary. One can solve a DP without recursion.
Memoization is a way to optimize DP algorithms which rely on recursion. The point is not to solve the sub problem again which has been already solved. You can view it as cache of solutions to sub problems.
