Accurate corneal layer boundary extraction from optical coherence tomograms can provide precise layer thickness measurements required in the analysis of corneal disease. This paper establishes a novel approach to precisely obtain the five primary corneal layer boundaries. The proposed method determines correspondence relationships between the layer boundaries to facilitate robust boundary extraction in the presence of noise and artifacts. The first phase of the method applies morphological operators to enhance the prominent structural features of the cornea. The second phase uses a semi-automated segmentation algorithm to extract the upper and lower boundaries of the cornea; these boundaries are used to register the corneal image. The final phase extracts all five boundaries using a global optimization method exploiting the medial correspondence relationship between each layers.The proposed method is tested and verified using a representative set of optical coherence tomography images and compared against several state of the art methods. The proposed method is demonstrated to be more robust to noise, to provide more accurate segmentation results, and to require fewer user interactions than the other published methods.