Single 2D Gabor wavelet vessel segmentation algorithm only considers the image filtering information, ignoring the vascular shape and structure information. A novel segmentation method based on 2D Gabor wavelet and combined line operator was proposed for extracting the retinal vessels more accurately and rapidly from retinal fundus images. Firstly, we construct a pixel feature vector consisting of the pixel intensity, four features from Gabor wavelet transform in different scales and one feature from combined line operators. Then, the segmentation of retinal blood vessels is implemented by the proposed feature vector in conjunction with a Bayesian classifier with gaussian mixture model. The proposed method has been tested with all retinal images in the publicly available DRIVE database. Experiment results show that the proposed method achieves an area under the receiver operating characteristic curve of 0. 9636 and accuracy of 0. 9486, it is distinctly better than the single 2D Gabor wavelet vessel segmentation algorithm.