Concerning the problem that the computational complexity of Coding Unit (CU) quad-tree splitting algorithm for High Efficient Video Coding (HEVC) intra prediction is very high, a fast intra CU splitting algorithm was proposed, which can narrow depth range of CU splitting. Firstly, multiple texture features were achieved through new defined texture extraction algorithm. Secondly, in order to get a decision function, Support Vector Machine (SVM) was employed to train the feature parameters. At last, based on decision function, CU splitting depth range was narrowed by skipping unnecessary splitting and early terminating splitting. The experimental results show that compared with the reference model HM 12.0, the fast CU splitting algorithm provides about 43.23% time savings on average with 0.84% increasing on bit-rate which obviously improves the coding efficiency. Besides, the proposed algorithm is easy to be combined with other methods to further reduce the computational complexity for HEVC intra coding.