High Quality Raster Image Resizing Using Linear Weighted Manipulation

Abstract

Due to the versatile customer applications and especially for large size industrial printing needs, the conflict comes between the image enlargement and high resolution quality. The image resizing process introduces specific pixels to the original image, or leaves out specific pixels from the original image. In order to maintain image quality and non-distortion, the value of such pixels and their positions were carefully deliberated. Prior researches such as DCT-based [2], wavelets, pixel replication linear interpolation [1] [3] those were limited by bandwidth and storage spaces, therefore a high efficiency resizing method is proposed in this paper. We introduce a new simple and high quality image-resizing algorithm that using linear weighted method for enlarging and drawing out particular pixels for reducing image size. The enlarging is first procedure and following with reducing procedure, where the image boundaries between those procedures needed to be carefully identification by linear weighted to avoid resolution loss. The simulation suggested the image can be enlarged up to four times and reduced down to quarter of original image at no resolution deterioration, and save the storage space need of fifty percent. Besides, the separation procedure is independent on enlarging procedure that can be operated if the image only needs to be reduced, same performance as mentioned before. This algorithm had been implemented by hardware and it will easily operate in software function in near future.