A binary image improving method based on bistable system

For the purpose of improving the quality of binary image and restraining its noise, a binary image improving method based on bistable stochastic resonance system was proposed in this paper. Firstly, it re-sampled the image to one-dimension signal according to the row or column and inputted the re-sampled signal to bistable stochastic resonance system. When solving the bistable formula by Euler method or Runge-Kutta method, it retrieved two partial differential coefficients along row and column. Then it calculated the value of neighborhood pixel with the coefficients according to the two dimension Talyor formula. For every pixel, it was calculated 8 times and then integrated to a value with different weights each times to retrieve a gray image. To segmenting the gray image to binary image, it got the final recover result. The experiment results showed that it recovered the heavy noise polluted image in effect and improved the PSNR(Peak Signal to Noise Ratio) 22.36% to the traditional wiener filter method and 10.8% to the contrast algorithm.