Abstract:
The Computational Photography field is gaining importance in consumer products mainly because of the applications proposed by the researchers in the field. The applications covers wide spectrum in digital imaging system, from the modifications at sensor level to the post processing the captured images. The further literature discusses about three such applications. The problem of image stylization using guided filter is studied for the first time. The pipeline proposed in the work shares some steps with previously proposed similar works but novel in the aspects of automation of user assisted segmentation and adaptive stopping criteria for iterative abstraction process. These adaptations in traditional abstraction flow removes user assistance completely. The comparison with other competitive algorithms and results of subjective analysis show the effectiveness of the proposed approach. Extending the HDR imaging framework and considering the complexities in HDR video generations, an intermediate solution in the form of a new computational photography application termed as 'HDR-GIF' is proposed. The proposed application is flexible in terms of the methods used in image alignment and HDR generation algorithm and can be thought of as a replacement for HDR video frameworks within system limitations. Further, one of the recent and popular approaches to extend the depth-of-field using coded aperture is discussed. Experiments are done to obtain calibrated kernels and all-in-focus image for the images captured with coded aperture. These two outputs can be combined to obtain depth map with improved resolution in future.