Pancreas is the organ that is responsible for regulating insulin levels. Damage to the pancreas causes diabetes mellitus. To find out if a person has diabetes mellitus, it is necessary to perform laboratory tests that are time consuming and costly. In this research, we propose a mobile application to check diabetes mellitus using iridology. Iridology is one of the alternative methods to detect abnormalities in the body through the iris. This study also aims to improve auto cropping from previous research to obtain higher accuracy. Stages performed in this research are cropping based on target, pre-processing, auto cropping using integral projection, auto cropping to remove sclera, retrieval of pancreas ROI, feature extraction and classification. The auto cropping shows the result of 56% success, 32% scant, 12% fail and and all the success cropping brings accuracy for classification.