Distortion Sensitive Algorithm to Preserve Line Structure Properties in Image Resampling

Abstract

In order to remove less important content from image seam carving algorithm is used. In seam carving distortion is very low as compared to other techniques like scaling and cropping. The major drawback of seam carving is when seam intersects with straight line present in the image it distorts line structure; the line may become curve after distortion. This structure distortion not only degrades visual quality of image but also gives artifacts or aliased line structure. This paper presents a content aware seam carving algorithm to resize the image. After applying algorithm discussed the structure of regular objects present in the image can be preserved. In the proposed algorithm first line detection algorithm is applied over the image in order to detect possible straight lines present in the image. After detecting straight lines algorithm tries to find out intersection point of optimal seam with the straight line. Algorithm increases energy of local neighbourhood pixels of intersection point up to a predefined radius, so that no further seam can intersect same pixel again.