Using a single image, manually add blur. The process of adding blur is that the image is convoluted with a function (PSF), and two blurred images can be obtained after convolution with two point spread functions with different parameters. First, the input of the system is a 640 * 480 image. After convolution with two gc-psf functions with different parameters, two defocused images are obtained. Then, the two defocused images are standardized to get the preliminary recovered scene depth map. Subsequently, we use the extinction La placian function to refine the interpolation.
(1) Defocus blur stage:
Fuzzy function used (point spread function)is ,and , 。
The variables are 、t。tThe value range of0~1，Usually take 0.1,0.2,0.3.....0.9 These nine values were used as the control of nine groups； can input and adjust the value of。The input image is one 640*480 grayscale image 。suppose t=0.3(theoretical best)，we have and Two different degrees of blur 和 ，Then use the input gray image respectively convolution with and ，then we can get the processed image and 。
(2) Image recovery stage：What we got from the previous step and make standardization，The specific formula is ，the IIs the initial grayscale image， is recovered images。In order to improve the quality of restored image， convolution with a 11*11 mean filter，then we can get
Using the method of extinction Laplace to refine the image, this step is realized in MATLAB
8 freelancers are bidding on average $529 for this job
Greetings! I am the professional electrrical engiener, verilog / vhdl expert, I have rich experiecne in communication mdoules, BPSk< OFDM, DDS, automatic control, image processign. more in caht. thanks.