Image corrupted vs Image not found vs File format not supported

you can do additional (manual) checks, like f = open(imgpath) for 2. or try to match the first bytes of the file against some 'file magic' list for 3. , but the real hard case is 1., since you'd need to know about the inner structure of the file format , you're trying to read.

the idea of try/catch won't work either, imho. you cannot "repair" a broken file on the fly.

edit flag offensive delete link

add a comment

answered 2018-03-14 04:14:02 -0600

Santhosh1
318 ●11 ●15 ●37

updated 2018-03-14 07:11:31 -0600

@berak Couldn't add this in the comment

I found code few code online using urllib and requests, which solves problem number 2

url='https://'
re=requests.get(url)
print(re.status_code)
if re.status_code == 200:
    resp = urllib.request.urlopen(url)
    image = np.asarray(bytearray(resp.read()),dtype="uint8")
    image = cv2.imdecode(image,cv2.IMREAD_COLOR)

Only part is how do I get the same None value for both corrupted as well as unsupported files that I use to get using imread

imdecode return empty matrix/image if the buffer is short or contains invalid data hopefully None in python

imread checks if the image cannot be read, then return same as decode empty martix and None in python

How to I add this improper permissions, unsupported or invalid format using python on numpy array of the image I already have?

Or is it not required?

edit flag offensive delete link

Comments

yes, that's what i would do, too. though:

Without using another type of library in python.

PS. iirc, this won't work anymore in python3, so careful !

berak ( 2018-03-14 04:20:34 -0600 )edit

I did check now, it does work fine. I'm using python 3.6 and OpenCV 3.4.0

Santhosh1 ( 2018-03-14 04:36:06 -0600 )edit

but please, screenshots of text are bad here, and you're refering to outdated 2.4 documentation

berak ( 2018-03-14 04:47:37 -0600 )edit

Sure, I did check it, those are from version 3.0.0. I did check with a PDF file its giving NONE worried about corrupted files now

Santhosh1 ( 2018-03-14 04:53:38 -0600 )edit

maybe just a link to current master docs will do ?

berak ( 2018-03-14 04:56:43 -0600 )edit

https://docs.opencv.org/3.0-beta/modu...

Santhosh1 ( 2018-03-14 05:03:30 -0600 )edit

https://docs.opencv.org/3.0-beta/modu...

Santhosh1 ( 2018-03-14 05:03:49 -0600 )edit

I can confirm, that imdecode is behaving the same way as imread behaves with the corrupted .jpg files and for .pdf for unsupported files. I'm getting None for both these cases

Santhosh1 ( 2018-03-14 05:26:03 -0600 )edit

yes, true, it behaves quite similar to imread.

also, no support for pdf in opencv at all. (it is not an image format, anyway)

still, as long as you have a raw byte buffer, you could check file magic, or things like 0xffd8 / 0xfdb for jpg

can we remove the outdated / misleading screenshots ?

berak ( 2018-03-14 05:34:44 -0600 )edit

add a comment

Image corrupted vs Image not found vs File format not supported

Comments

2 answers

Comments

Links

Question Tools

Stats

Related questions

Image corrupted vs Image not found vs File format not supported edit

Comments

2 answers

Comments

Links

Question Tools

Stats

Related questions

Image corrupted vs Image not found vs File format not supported