<p>I have a use case where I have to match images even pixel by pixel will do fine, shall I iterate through each pixel and check if first one is equal to the second one?</p>
<hr/>**评论：**<br/><br/>dlsniper: <pre><p>It depends on what you want to achieve. If you want to check that they are pixel identical, then yes. But the same image might be just a bit more different than the other because of a different compression setting/format. There really is no good way to answer without more details and even then it might not be easy to implement (if at all possible). What happens if the images are on a different resolution? </p></pre>thewhitetulip: <pre><p>Yes I agree to your concerns about resolutions, but in the scenario in which I am to work that is plain and simple, say if someone uploads the same image ten times, then we have to say that there are 9 duplicates</p></pre>dlsniper: <pre><p>Then why not simply crc32 them and see if the hashes match? Or some other fast checksum such as xxhash (if that works)? </p></pre>TheMerovius: <pre><p>The standard go image types (for example <a href="https://godoc.org/image#NRGBA" rel="nofollow">NRGBA</a>) export Pix, an array of bytes. Using <code>bytes.Compare</code> of those should be considerably faster.</p></pre>TheMerovius: <pre><p>Also, if you're after byte-by-byte duplicates you should store a hash of Pix along with your Image in your database and look for that hash instead. Otherwise this won't even remotely scale.</p></pre>lkjhgfds2: <pre><p>pHash is possibly what you want here. <a href="https://www.reddit.com/r/golang/comments/2wnns5/findimagedupes_use_phash_to_find_visually_similar/" rel="nofollow">There was a thread</a> about it some time ago.</p></pre>
