最後,本文結合顏色分塊和Bilinear 以及顏色交界處的自定義插值法,只需要增加少量的處理時間,便能得到比Bilinear 較佳的視覺效果。 In present time, many consumer electronic devices have build-in digital camera that makes peoples have more chances to use digital camera related devices. Therefore, digital image processing has become an important research topic in present days. The technique of image scaling is always a major topic of the digital image processing, especially the image upsampling technique. In this thesis, by using the color information of the image, a bilinear interpolation based method is proposed for smoothing the edge of the objects in the image.
In this thesis, a novel method is proposed to segment the image according to the color information at first. The segmentation process starts from the border of the image, and assemble the pixels that have similar color. In this work, the image is cut into several blocks. When the small block contains features of the color edges, the upsampling process will be performed on the corresponding block by the proposed method based on the bilinear interpolation technique. Otherwise, the upsampling process is performed using the pure bilinear interpolation technique.
The proposed method combines the color information and modified bilinear interpolation technique to achieve better visual quality. Moreover, comparing to the traditional methods, the proposed method only requires a few extra computational time to accomplish the goal.