Seam Carving Approach for Multirate Processing of Digital Images

Abstract: – This paper presents a new approach called Seam Carving for multirate signal processing of digital images. Increasing and decreasing the sizes of digital images are common place in day-to-day image processing. These methods involve long procedures and sometimes consume more time for getting implemented. Whereas, using the Seam Carving method eradicates the excess time involved in upsampling and downsampling a digital image considerably as this method is straightforward and simple to implement. This method comes in handy when we are dealing with large medical images and remotely sensed images. This technique is applied on standard images and its performance is analyzed. The entire work was implemented using Matlab R2017a software package.

Keywords – seam, image, upsample, downsample.

INTRODUCTION eam Carving method is one of the latest digital image processing techniques that are used for expanding and decimating the digital images. It is essentially a content-aware image resizing algorithm that was invented by Shai Avidan and Ariel Shamir in 2007 [9]. The seam carving algorithm is also called Liquid Rescaling algorithm. Since images, including bitonal and colour, are multidimensional, Seam Carving algorithm is also essentially a multidimensional digital image processing algorithm that is applied in both the spatial and frequency domains. Seam Carving method works on the principle of computing the paths of least importance referred to as seams in a digital image. Expanding an image comprises of seams being inserted into it whereas, decimating an image comprises of seams being eradicated from it respectively [10]. The liquid rescaling method enables manually defined arenas in which pels may not be modified, and features the ability to eradicate objects in entirety from the digital image. This method is used for properly displaying images without any distortion on different projection media.
Read More