How to extract image from TCP Stream

Question

I need your help.

I was creating an application in c# that converts the data from the IP camera to an image (JPEG).

I was able to convert the image using the below code:

 hex = "FFD8FFDB008400130D0F1.........";/// supply this with the attached hex dump.
 byte[] image = HexString2Bytes(hex);
 File.WriteAllBytes("visio.png", image);
 Process.Start("visio.png");

private static byte[] HexString2Bytes(string hexString)
{
    int bytesCount = (hexString.Length) / 2;
    byte[] bytes = new byte[bytesCount];
    for (int x = 0; x < bytesCount; ++x)
    {
        bytes[x] = Convert.ToByte(hexString.Substring(x * 2, 2), 16);
    }

    return bytes;
}

Sometimes I get a better image as expected:https://ibb.co/pxrwn6p

but sometimes I get a distorted image after converting https://ibb.co/9twx5ZT.

I was wondering if there is a problem with the conversion or the way I save the image. because as per the supplier what I need to do is to directly save the image from the stream. but since I receive it as a byte and I still need to convert it maybe there is something wrong with my codes. the image also starts with ÿØÿÛ FF D8 and ends with ÿ Ùÿÿÿÿ (FF D9 FF FF FF FF)

here's the hex dump from their sample app: https://drive.google.com/file/d/1CMlQ0xaVjM0jfU5A4MB-_HwK54dUMTOr/view?usp=sharing

using their test application the image can be captured and converted the image perfectly. captured image using their application:https://ibb.co/2KgyLTc

using the hex from the sniff and convert it using my code: converted image using my code:https://ibb.co/G0WMjht

sample source code: please bare with my codes because currently this is only my test app before integrating this feature to another app. https://drive.google.com/file/d/1Ux7zsR39IVNyd1wrBxQPQKA6yM4YnwJN/view?usp=sharing

Thank You in advance.

Answer 1

Looking at the hex-dump it looks like some kind of XML file with embedded image data. Trying to convert this directly to an image will most likely not work, you would need to parse the XML-data to extract the actual image file. But it looks like you have a valid Jpeg header, so I would guess you have found the start of the image at least. But you probably also need to check the length property from the XML-data to find the length of the image-data block.

However, the datablock looks like it contains large sections of zeros, this should not be present in a jpeg file, so it might indicate some data corruption. Possibly from the way the network data is captured.

I would expect cameras to use some higher level protocol than raw TCP. Like Real Time Streaming Protocol, GigE vision, or mjpeg over http. I have not seen any camera that require you to process a raw TCP streams. But since you do not show how the data is fetched it is difficult to tell if there is any mistakes in that code.