Research On Template Selection Of Inverse Halftoning Based On Look-up Table

Halftone technique turned the continuous tone images to two value images, greatly reduces the image reproduction cost in printing and display on the device. If enhancement, compression, zoom operation should be operated on the halftone image, halftone image should have been inverse halftoned first. Look-up table(LUT) inverse halftoning technology, with its advantages of simple calculation, low complexity,parallel processing and easy programming, has gradually become one of the most important current inverse halftoning.This in-depth study aiming at the problem of template selection which effects LUT inverse halftoning quality. The main research work is as follows :1) Proposed an optimal LUT template selection method based on genetic algorithm. In view of the problem that greedy algorithm is easy to fall into local optimal solution in the process of finding LUT template,this paper presents a method to select optimal template based on genetic algorithm. The algorithm has high efficiency and global convergence and strong parallel processing ability etc.. The obtained LUT template greatly improved the peak signal-to-noise ratio of the inverse halftoning image.2) Proposed a hybrid optimal LUT template selection method, based on the integration of genetic algorithm and simulated annealing algorithm.First, for the shortcomings that genetic algorithm is prone to "premature",this paper puts forward a method of LUT template selection based on simulated annealing algorithm and improved the traditional simulated annealing algorithm temperature update mode. On the basis of the fusion that GA algorithm has high whole searching ability and SA algorithm has fine local search, this paper puts forward a hybrid optimal template selection algorithm. Hybrid algorithm, when searching the optimal LUT templates, spends longer time than individually using GA, SA algorithm,but the LUT template found by hybrid algorithm has higher quality, and it has more stable performance and stronger global optimization ability.3) Proposed an improved LUT inverse halftoning algorithm for color image based on the color channel related. Traditional LUT inverse halftoning algorithm for color image is improved, extended the traditional single channel template to three channel, fully considered the correlation between color image’s R, G, B three channel. And applied the hybrid optimal template selection algorithm to color image LUT inverse halftoning, to find the optimal LUT template. Through the experiments,it proved that the color channels relatived LUT inverse halftoning algorithm is better than the traditional color inverse halftoning algorithm in the objective and the human visual effect is better, and color reproduction is more realistic.