Summary: | 碩士 === 國立臺灣大學 === 資訊工程學研究所 === 87 === Digital zoom is the process that to scale up a digital image to another
higher-resolution image by using a computer. By using digital zoom technique, an
image taken from a low-price and low-resolution camera can be scaled up to a
higher-resolution image. Even if we have a high-price and high-resolution
camera, there exists the resolution limit. At this time, we can also observe
details in an image by applying digital zoom algorithms. Furthermore, the image
obtained from the network or the scanner can be also scaled up by using digital
zoom. Traditionally, digital zoom can be achieved by interpolating the
neighboring pixels, and bilinear interpolation is one simple and fast method
among them. However, the drawbacks of bilinear interpolation are the blocky and
blurry effects on edges in a scale-up image. Therefore, we propose ''Adaptive
digital zoom techniques based on hypothesized boundary" to avoid these effects.
Based on the hypothesized boundary concept, we develop four algorithms according
to the interpolating pixels'' distance to nearest run boundary. These four
algorithms are linear weighted sum, sigmoid weighted sum, Larger-NRBD selection,
and area-based restoration and resampling. The experimental results show that
our approaches are better than bilinear interpolation.
|