A hybrid approach for image super-resolution of low depth of field images

Multimedia content plays significant role in our day to day life. Nowadays devices like mobile, tablets, Digital TV have capability to display high resolution images for better viewing experience. Therefore the need arises to enhance the resolution of available images in real time for better perceptual quality. This paper focuses on super-resolution of shallow depth of field images which are widely used in macro, portrait or sports photography. The proposed hybrid approach first segments the low resolution image into the object of interest (foreground) and background. Subsequently, super resolve the object of interest by sparse representation and background by traditional interpolation approaches. This approach helps to reduce the overall time complexity and enhance the quality by extracting significant features from object of interest and reduce computations for shallow depth of field images. Experimentally it has been inferred that hybrid approach performs better than existing state-of-the-art super-resolution approaches in terms of time complexity.