The theory of transform and quantization in H.264 is introduced in detail. The complexity of transform is lower byusing a 4×4 integer transform

which is derived from DCT transform. And the mismatch problem resulted by floatcomputation is also avoided. A scaling multiplication （part of the complete transform） is integrated into the quantizerto reduce the total number of multiplications

which makes H.264 an effective video coding standard. At the end

weadopted SSE2 Intel instructions to implement transform and quantization for H.264. The experiment shows that theproposed algorithm can improve coding speed greatly.