1.7 Results and conclusion

Results and Conclusion

In the noise analysis, we compute the $\text{PSNR}$ as a function of noise variance for a variance range from 0 to 0.0005. We do this for thedeblur results of both the flutter shutter blurred image as well as the boxcar filter blurred image. We plot these results, which you can see infigure below. Here, the $\text{PSNR}$ characterizes the deblur result quality and the noise variance characterizes noise level. Two importantresults and conclusions can be drawn here.

Flutter Shutter: The flutter shutter deblur results always have a higher $\text{PSNR}$ than the boxcar filter deblur results at all noise variances. This shows that images taken with a flutter shutter cameracan be deblurred more successfully than images taken with a traditional boxcar filter in all cases of noise.

Flutter shutter deblur - noise variance 0

Flutter shutter deblur - noise variance 0.0005

Boxcar: The boxcar filter deblur result's $\text{PSNR}$ decays extremely quickly to a minimum value of 5 even with only a small amount of noisein the boxcar filter blurred image. On the other hand, the flutter shutter deblur result's $\text{PSNR}$ decays rather gradually with increasing noise in the flutter shutter blurred image. This resultshows that flutter shutter images and their deblur results are more resistant to noise damage than are boxcar filter images. Deblurredpictures demonstrate this conclusion, showing that even at a noise variance of only 0.0005, the boxcar filter deblur result is alreadyheavily damaged while the flutter shutter deblur result quality is maintained comparatively well. .

Boxcar deblur noise variance 0

Boxcar deblur noise variance 0.0005