I am developing an adaptive thresholding method based on integral sum images and sauvola method. I need a dataset to test my method against the ground truth. any help is appreciated.
Your image is distorted and blurred. Before applying any threshold and segmentation, you shoul
i) normalize/rectify it, and
ii) remove the blurr.
Once you preprocessed your data, then the rest becomes easier.
A common image processing problem is to recognize digits and letters. Depending on the "alphabet" you have, you may not need a ground truth.
If you know the source of your "alphabet," then you can either make your ground truth or look at the references related to it.
here's an example of what I am trying to threshold. so the purpose is to threshold this image so that I can get the black squares clearly and well separated from one another.
i hope content of this link can be helpful for your task!!
Chapter Ground Truth Data, Content, Metrics, and Analysis
Your image is distorted and blurred. Before applying any threshold and segmentation, you shoul
i) normalize/rectify it, and
ii) remove the blurr.
Once you preprocessed your data, then the rest becomes easier.
A common image processing problem is to recognize digits and letters. Depending on the "alphabet" you have, you may not need a ground truth.
If you know the source of your "alphabet," then you can either make your ground truth or look at the references related to it.
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing