Research Of Image Interpolation Algorithm

Image interpolation plays an important role in image processing area, and it is widely used in various fields. Traditional interpolation algorithms are simple and easily realized. The difference between traditional algorithms is the way of getting the pixels around. Though traditional algorithms achieve good visual effect in flat ares, but they can't process the edges, textures and so on, which leads the edge details to be fuzzy in the enlargement image.Because of the subjective evaluation of image quality is mainly affected by the edge, some scholars propose many edge-directed interpolation algorithms. This paper analysis kinds of methods popular nowadays, and finds out that many interpolation algorithms which obtain bigger promotion in precision, actually need more complex computation.Data fusion technology that through the combination of information on different directions to gain more accurate information than information only on a single direction which not only get good vision effect, but also cost little computation. So, in order to improve instantaneity of algorithms, many scholars start to research on algorithms based on data fusion technology. Nowadays, though some adaptive interpolation algorithms based on fusion technology get a breakthrough on interpolation speed, the consideration of fusion method is still insufficient. Thus, this paper proposes an improved algorithm based on orientation-fused. Firstly, region type judgment, bilinear algorithm used in flat area; improved algorithm used in edge area. Improved algorithm fully considers six estimate results of horizontal, vertical and diagonal directions, and then utilizes the vertical distance and direction gradient to calculate weights so as to obtain a final result. Through a large number of experimental data and renderings, we can make sure that proposed algorithm is feasible and efficient. The improved fusion algorithm based on multiple directions can make edges of amplificated image more natural and clear, also can achieve better texture effect, cost less computation time which greatly enhance instantaneity than the other algorithms.When interpolating a color image, the key point is to choose a good color space. In this paper, we select HSI color space, which is visual perceptional and equipment-independent.We also do some research on color interpolation algorithm based on pictures got form CCD. In order to reduce cost and complexity, digital camera manufactures usually employ a single CCD sensor whose surface is covered with a color filter array, named as CFA. The CFA limits each sensor pixel just sampling one of the three primary color values. In order to get a full color image, there must be an interpolation algorithm designed to recover the two missing colors. This part looks back some typical algorithms, then, analyses advantages and shortcomings of the typical and part of the new algorithms. At last, a new interpolation algorithm based on the weight of directions is proposed. New algorithm divides images into strong-edge-region, weak-edge-region and flat area, and fully uses chromatism criterion and weights of multiple directions in the process of recovering missing channels'information. The proposed algorithm based on weights of directions can make the image reconstruction get better visual effects, higher average gradient value, smaller Euclidian distance and cost less computation time than the other algorithms.