COMPUTATIONAL FILTER-APERTURE APPROACH FOR SINGLE-VIEW MULTI- FOCUSING Vivek Maik, Dohee Cho, Sangjin Kim, Donghwan Har and Joonki Paik Graduate School of Advanced Imaging Science Multimedia and Film, Chung-Ang University, Seoul 156 756, South Korea. (vivek5681@wm.cau.ac.kr ) ABSTRACT Most of the focusing techniques need to estimate depth information for ensuring that the object of interest is at an appropriate distance for full frontal focus. Computational cameras which can variably focus different regions of the scene with large depth of field have been proposed. In this paper we propose a full auto-focusing algorithm using computational camera without involving any digital image restoration methods and just one input. The proposed computational camera uses multiple filter apertures corresponding to each color channel which can acquire three shifted views of a scene in the RGB color planes. We can make any region focused by appropriately shifting each color channel to be aligned. Depth map estimation is carried out to extract different regions from these channel shifted images which is later fused to produce the final image without any focal blur. Experimental results show performance and feasibility of the proposed algorithm for auto- focusing images with one or more differently out-of-focused objects. Index Terms— Image restoration, image classification. 1. INTRODUCTION Demand for digital auto-focusing techniques is rapidly increasing in many visual applications, such as camcorders, digital cameras, and video surveillance systems. Conventional cameras have come a long way in dealing with problems associated with focal settings and blur. Even though several steps have been taken, focal blur caused by varying distance of the object from the lens has been something that the conventional cameras still have to deal with. With focus set at either near, mid or far regions of the scene, the captured image tend to have only that particular region in focus where as the remaining regions tend to be in out-of-focus. Post- processing steps in the form of blur restoration and multiple image fusion have been proposed to deal with the focusing problem. Recently computational cameras have been developed that were capable of capturing additional information from the scene which when combined with post-processing can overcome several drawbacks of the imaging applications including: refocusing, increased dynamic range, depth-guided editing, variable lighting and reflectance, etc. The scope of this paper is to deal with the first factor that is the focal blur due to the mismatch in the distance of the object and focal length of the lens. In this paper we propose a combined hardware-software approach by which we tend to model the focal blur as channel dependent depth maps which are then used to remove the focal blur in images. The former (hardware) refers to the computational camera which employs a novel multiple filter aperture (FA) models for separating and distributing the blur into different color channels as shown in Fig. 1. Fig. 1. (a) The modified camera configuration with the proposed multiple color filter-aperture. A normal aperture in a traditional camera, shown in (b), is replaced with the modified aperture consisting of red (R), green (G), and blue (B) color filters as shown in (c). The latter (software) refers to the computation or algorithm part which involves variable focusing, depth map estimation and fusion to generate a fully focused image from just single FA input. The block diagram of the proposed single view auto focusing algorithm is shown in Fig. 2. Fig. 2. (a) Block diagram of the proposed algorithm using computational camera, (b) five objects including a bunny(orange), a flower (red), a star (yellow), a tree (green), and a cloud(sapphire green) located at different distances from the camera, (c)-(f) captured image and respective color channels, (f-k) four differently focused images with optimal focusing at the bunny, the flower, the tree, and the cloud, respectively and channel depth. 1541 978-1-4244-5654-3/09/$26.00 ©2009 IEEE ICIP 2009