Dear all,
I am not an expert in image processing. I have an set of jpeg images which I need to extend its size artificially, for example, add some noise, change luminance, chrominance etcc... how can I do that? is it possibile to generate an artifical dataset out of what I have?
Thank you,
Kassem
To work on specific images you can use ImageJ to filter them, add noise etc. If you want to generate images, one option is to use this: http://www.i-tex.de/news.php
As Nek suggests, imageJ will let you add Gaussian blurs etc to increase your panel of images. I would add a word of caution though - standard jpeg is a lossy compression format, which means each time you save it you will throw away more of the original information, producing significant artefacts. Generally, in image analysis, jpegs should be avoided at all costs. If you can then stick to a non-compressed tiff or similar to save your image data.
Hi Kassem,
Another option is to extend your set of images by generating batches of tensor images. You can do that easily using Keras ImageDataGenerator. Take a look at this link:
https://keras.io/preprocessing/image/
Best Regards,
Heitor
- Badges
- Science topic
- Similar topics
- Computer Science
- Images