Abstract: The most prominent feature of intra prediction in the High Efficiency Video Coding (HEVC) is adopting the tree structured coding unit (CU) , whose size can be changed from 64×64 to 8×8, and 35 intra prediction modes. Consequently, the intra mode decision process in HEVC is performed using all the possible CU sizes and prediction modes to find the optimal coding mode. This achieves the highest coding efficiency but at the cost of a very high computational complexity. To reduce the computational complexity of intra frame prediction, from the version 2.0, the HM reference software has adopted an two stage fast intra prediction algorithm. In this paper, an improved fast mode decision algorithm for intra prediction is proposed based on two stage fast intra prediction algorithm to further reduce its computational complexity. Firstly, a rough decision based on the texture direction is exploited to reduce the computational loads of rough mode dicision. Furthermore a refine decision method based structure correlation is proposed to reduce the number of candidates in the rate distortion calculation process by using texture similarity in a homogenous region in order to further reduce the computational loads of intra prediction. Experimental results have demonstrated that compared with the fast intra mode decision method of HM9.0, the proposed algorithm can save 25% computational time on average with negligible loss of coding efficiency.