Motion Deblurring From A Single Image

During the camera’s exposure time,the image can be blurred because the relative movement between the camera and the object being photographed.Sometimes,it is impossible to take a photograph again or the cost is expensive.This type of motion blurred image is useless and worthless.Motion blurred images are very common in life。Camera shooting in the shaking hand and taking pictures on mobile devices can cause image blurred.Restoration technology for motion blurred image is applied to many fields,transportation,aerospace,criminal investigation and military fields are involved.Because of the advancement of science and technology,the application of images is an important way for human to obtain information in various fields.Motion blurred image is difficult to recognize.In order to solve this problem,many scholars have done a lot of work and have remarkable results.Single image lack information about restoring blurred image,so it is difficult to restore burred images.For the blurred images caused by linear motion,blurred parameters include blurred length and blurred angle.It can determine the point spread function.The point spread function is the key step in motion blurred image restoration,and its accuracy is directly related to the quality of restoration.Based on the characteristics of blurred images caused by linear motion,a degradation model is established.According to the Fourier transform of the degraded image,the fuzzy parameters are identified.We deduce the relationship among the fuzzy parameters,spectral stripes and image size.In this paper,the fuzzy parameter identification method is improved.The first cut is made according to the length and width of the spectrum.The contrast of the spectrum is enhanced by histogram equalization.After Canny detection,we take the spectrum image center as the center point and search the edges in the four directions.The range of the second interception is determined from the distance between the searched point and the center point.The Hough line detection is performed on the second intercepted image.In this way,the range of the clear region can be obtained automatically according to the image size and the fuzzy length,which increases the precision for extracting blurred parameter and reduces the calculation amount.After the fuzzy length and fuzzy angle are identified,degraded images are restored by linear and non-linear restoration methods respectively.We compare the two methods in terms of time and recovery effect.During the restoration process,because the boundaries of the image are not continuous,it cause the ringing phenomenon.In this paper,the optimal window method is used to suppress the ringing phenomenon.