Anterior chamber (AC) angle is a useful parameter in the investigation and diagnoses of angle closure glaucoma. And Optical coherence tomography (OCT) is a noninvasive and high-resolution technique for AC angle imaging. An accurate and automatic angle measurement should be based on a precise AC angle segmentation in OCT image. However, conventional segmentation algorithms can not achieve a well result because the algorithms are not robust to the dark shadow areas and speckle noise existing in the image. A new compound segmentation algorithm is proposed in this paper. The morphological method, intensity transform, gradient image, and ratio of foreground-to-background average intensity, etc., are applied in this new algorithm to improve the accuracy of segmentation. The experiments show that the presented algorithm can achieve a satisfactory segmentation result for the AC angle in OCT images.