You need either to create separate rules for all tags or (provided source and target are filled with text) process xlf as xml and add html parser for embedded content, where you can declare, that parts of the img tag shall be translatable. If your xlf has just source and empty target, you can also open such xlf in Studio, copy all source to target and save as target. This way you get a xlf with source and target filled with the same text.

However, provided your information, the best and easiest way here would be creating a proper parsing rules for the tags in the xlf.

You need either to create separate rules for all tags or (provided source and target are filled with text) process xlf as xml and add html parser for embedded content, where you can declare, that parts of the img tag shall be translatable. If your xlf has just source and empty target, you can also open such xlf in Studio, copy all source to target and save as target. This way you get a xlf with source and target filled with the same text.

However, provided your information, the best and easiest way here would be creating a proper parsing rules for the tags in the xlf.

Jerzy Czopik said:

the best and easiest way here would be creating a proper parsing rules

Thank you Jerzy Czopik for your input. 
I understand that I need a rule. But I can't figure out what rule. Do you mean I should remove the default general rule and add individual rules for each type of tag? Actually, I thought of that method too, but I just hoped that there could be an easier way like a regex phrase within the existing default rule that can just exclude 'img alt' tags from that general rule. However you seem to be right: I have to add separate rules. Ok, thank you.