Tweet
DSP effect for Image Converter to detect blur amount & downsize the image till sharp
Many images, especially those taken by cell phones, are not that sharp and so waste space and give poor results when Ai up-scaling.
It would be useful to implement a DSP effect that analyses the images for blur/out-of-focus amount, then calculates optimum pixel dimensions to result in a sharp image and use that data to downscale to the sharpest result based on a threshold of total results set by the user. Could have an option to keep aspect ratio or not depending on the type of blur.
There are multiple techniques to do this, so I've posted what I've found below:
Discussion on stackoverflow:
Link to a paper outlining various techniques:
Analysis of focus measure operators for shape-from-focus
Using Wavelet Transform:
Blur Detection for Digital Images Using Wavelet Transform
Using opencv-python:
Blur Detection with opencv-python
It would be useful to implement a DSP effect that analyses the images for blur/out-of-focus amount, then calculates optimum pixel dimensions to result in a sharp image and use that data to downscale to the sharpest result based on a threshold of total results set by the user. Could have an option to keep aspect ratio or not depending on the type of blur.
There are multiple techniques to do this, so I've posted what I've found below:
Discussion on stackoverflow:
Link to a paper outlining various techniques:
Analysis of focus measure operators for shape-from-focus
Using Wavelet Transform:
Blur Detection for Digital Images Using Wavelet Transform
Using opencv-python:
Blur Detection with opencv-python