High Precision Image Rotation Using DCT Transformation

Human increasing need image processing with the development of science and technology. Image processing involve in human everyday-life, like photo processing of beauty, taking photo of traffic violation in high-speed or complicated environment, medical CT images and remote sensing image processing. Because of image-processing application in wide range and high quality, image processing develop rapidly. Images interpolation is a basis of image processing technology, which make high resolution technology be used in range of area, like improving image resolution and definition, geometric correction of remote sensing image, image rebuild, after effect of media advertisement and so on. Images interpolation is a process of data regeneration, and differential process means calculate(via some algorithm or function) the unknown integral-point between sampling points. The paper focus on the interpolation technology of image rotation. In this paper, we presented a new way to calculate the interpolation of image rotation, and the innovation can make image from time domain swap to frequency domain via discrete cosine transform. The pixel value of relevant coordinates can be result from analysing frequency of unknown coordinate points and processing inverse transformation after last stage.We already proved this interpolation method of image rotation more accuracy and stable compare with other three interpolation methods in this paper(the nearest interpolation method, bilinear interpolation method, bicubic and Hermite interpolation).