On analyzing various density functions of local binary patterns for optic disc segmentation

In building an automated glaucoma detection system, optic disc segmentation is the first step that needs to be implemented follows by optic cup segmentation in order to quantify the severity level of glaucoma. Glaucoma is an ocular eye disease that can lead to gradual vision loss and permanent blindness if it is not treated in the early stage. Many glaucoma patients are unaware of their disease since they rarely encounter any symptom that can lead to glaucoma. Thus, detecting glaucoma during the early stage is very important to reduce the treatment risk. This paper proposes optic disc segmentation by using local binary patterns operator (LBP), a feature for textural classification in image processing. LBP is utilized only on red channel of RGB fundus image because of higher contrast between optic disc and its surrounding area compared to the blue and green channels. Smoothing technique, specifically, histogram equalization is performed to improve the quality of input image before LBP method is applied. Lastly, morphological operation and filtering are applied to filter out the artifacts and remove the noise from the segmented image. RIM-One database is used to validate the simulation results with Exponential distribution achieve better performance with average accuracy and precision of 0.8951 and 0.7390 respectively.