Research Of Fast Image Interpolation Algorithm On Mobile Terminal

Nowsdays, with the development of cache, CPU and memory, the processing capacity of mobile terminal has been increased. Meanwhile, because of the popularity of the 4G network and the decrease of the mobile network charges, the mobile internet has been the focus of the world. The multi-media technology based on mobile internet, like video monitor and broadcast, is generally nourishing our life. To adapt to the high resolution of various multi-media terminals, the video display algorithm is needed to decode, amplify every frame of the video and show it on the screen. The essential features of these video display applications are smooth display and good user experiment.This paper focus on image interpolation. After the Comparison with the traditional interpolation algorithms on interpolation quality and time complexity, this paper raises a fast image interpolation algorithm which is adapted to mobile terminal. And then uses NEON technology to parallelize under the ARM platform.First, this paper introduces several interpolation algorithms like nearest neighbor interpolation, bilinear interpolation, bicubic interpolation, ideal sine interpolation, B spline interpolation and Lanczos interpolation.This paper shows the result of the contrast experiments between these algorithms and analysis their interpolation quality and time complexity. The result shows that the bicubic interpolation algorithm is more suitable for image interpolation for mobile terminal.Second, inspired by Zhe Wei, this paper raises a fast image interpolation algorithm for mobile terminal and program it. Unlike the image interpolation algorithm based on edges, the algorithm the paper represents not only takes the position of edges into consideration, but also quantity the edge intensity. The algorithm diffuses the original edge based on the different edge intensity and then generations new edges. The algorithm stores the edge information during the interpolation process and sharpen the interpolation image, strengthen its edge.Third, this paper applies the proposed algorithm to the laboratory research project, and uses NEON instructions to parallelize the algorithm based on ARM structure. The experiment result shows that the interpolation quality of this algorithm is higher than the traditional interpolation algorithm. Its cost time is about 2-3 times of bicubic interpolation and lower than B spline interpolation. With the support of the NEON technology, the average efficient of this algorithm has been raised by 80%-120%. Its efficiency is close to bicubic interpolation. It can be use to display high definition videos on mobile terminals smoothly.