Kaveh Pasandideh, I seem to have misunderstood the situation. The figure is not what you plotted from the raw data, but was obtained from another place as this png image (with axes and numbers). Then, the true value of the pixels on the axes and numbers are unknown, right?
I suggest contacting the original data holder, i.e., the corresponding author of the paper, and requesting that they provide the values. I believe this is the only correct way.
It is possible to estimate the values of areas hidden by axes, etc., from the surrounding pixel values; drawing software such as Photoshop can help you with this task (like removing a mole from a mug shot). However, the values created are not fair. It will be a creation of the data, not a recovery. I worry that you will be confused in the presentation when asked for the scientific evidence.
Kengo Shibuya thanks you very much. yes i have done some modifications on the pics and it seems that some data will lose. So i contacted with the data holders and they gave me the access to needed data.
Kaveh Pasandideh, OK, I hope you will do so. The provision of data is both a service and an obligation of the authors to their readers. Good luck with your research.