Research On Digital Image Non-periodic Importance Sampling Technology

With the development of computer technology,as well as the growing popularity of digital devices,people could easily acquire lots of high-precision digital images,which making digital images becoming a extremely important form of digital media.In the field of educational technology,the application of digital images has become wider in multimedia assisted teaching.Due to the higher increasingly fineness of the digital images and the bigger growing volume data,so that the digital images have difficultly in the process of storage and network transmission,which obstruct the using and spreading of digital images.Therefore,by using the sampling processing with the digital images,the volume of the digital images is reduced to meet the actual demand.To study the existing sampling technology,we propose an aperiodic-based importance sampling technology with blue noise properties.The main contents include constructing the aperiodic tilings,subdividing adaptivlity based on the digital image and optimizing the sampling points.The major contributions of this paper are:1.The generation algorithm of aperiodic tilings are proposed.As a special kind of tiling,aperiodic tiling has extremely powerful irregularities,so it is difficult to find a common way to construct the aperiodic tiling.By studying Penrose Tiling and P-type domino structure,we put forward an algorithm to quickly generate aperiodic tiling.In this algorithm,we firstly determine the rules to generate aperiodic tiling,then according to the rules,we subdivide the aperiodic polygons which contain in the algorithm,and zoom all aperiodic polygons to the original size.Through much subdivision,we may get the aperiodic tilings which contain the specified number and size.2.We present an importance sampling algorithm of digital image basing on the aperiodic tilings.The algorithm can subdivide the aperiodic tilings according to the texture features of digital image,and then generate the corresponding sampling points in the formation of aperidic image after the subdivision.This algorithm controls the subdivision of the aperiodic graphics through the texture features of digital image,which could be applied to any digital images with different texture features.Moreover,We can control the degree of subdivision to control the number of sampling points,in order to obtain the sampling points with different accuracy.3.Noise reduction processing is performed on the sampling points.In the process of sampling,it would produce a certain noise,which affects the final effect.In order to convert the correlated noise that was generated in sampling process into uncorrelated noise,we can set a sampling density threshold for each sampling point to judge whether the sampling point need to retain,thus filter the noise generated by sampling progress.The sampling points are fixed position in the aperiodic polygon,so the sampling point set distribution effect is still not ideal,we must correcte the location of sampling points,so as to improve the balance of the spatial distribution of sampling points.