Hi,
You will not be able to get what you need with just Keyword tests.
Exact implementation of what you'd like to get requires optical character recognition (OCR) for the picture file and intelligent match of recognized data to JSON content. For OCR you may consider either old native functionality provided by TestComplete (https://support.smartbear.com/testcomplete/docs/testing-with/deprecated/ocr/index.html) or the new one provided via Google API (https://support.smartbear.com/testcomplete/docs/testing-with/object-identification/ocr/index.html).
Implementation does not seem to be trivial, so maybe you should discuss possible alternatives that are more simple in implementation but covers the same verification.
