Image quality analysis of segmented iris using filters

Abstract

Quality of an Image has a significant impact on the overall performance of an image processing system. In this technical era, when everything is being digitized, many of us prefer to transfer data in a digital form; to be more secure while using digital devices through biometric sensors. We encounter many more such instances in our day to day life where these devices and sensors are deployed. The biometric devices or systems require a quality input to achieve quality performance. There is always a need to enhance the quality of the sample after its acquisition. In this paper, we are going to discuss image quality enhancement for biometric recognition system where iris samples have been used; by proposing an image quality analysis approach of segmented irises based on quality filters. It means input images are not directly passed to the filters, they are firstly segmented and then segmented parts will be enhanced using different filters. Segmentation is one of the initial key steps of recognition systems and has a significant role in the biometric systems as the performance results of the system’s subsequent stages depend upon the segmentation results. We have segmented iris images into three major parts namely ROI, pupil and sclera. ROI represents the region of interest i.e. iris. Moreover, iris images from four directions viz. up, straight, left and right are taken into consideration to achieve more promising results. Unlike general case, filters are applied for enhancement after segmenting the input images. To measure the quality of input samples, various image quality metrics have been calculated and analyzed comprehensively to reach to the conclusion that Gaussian filter performs better as compared to Average and Circular Average Filter.