How to hash only image data in a jpg file with dotnet?

Question

I have a ~20000 jpg images, some of which are duplicates. Unfortunately, some files have been been tagged with EXIF metadata, so a simple file hash cannot identify the duplicated one.

I am attempting to create a Powershell script to process these, but can find no way to extract only the bitmap data.

The system.drawing.bitmap can only return a bitmap object, not bytes. There's a GetHash() function, but it apparently acts on the whole file.

How can I hash these files in a way that the EXIF information is excluded? I'd prefer to avoid external dependencies if possible.

Answer 1

You can load the JPEG into a System.Drawing.Image and use it's GetHashCode method

using (var image = Image.FromFile("a.jpg"))
    return image.GetHashCode();

To get the bytes you can

using (var image = Image.FromFile("a.jpg"))
using (var output = new MemoryStream())
{
    image.Save(output, ImageFormat.Bmp);
    return output.ToArray();
}

Answer 2

Translating to powershell, I get this -

[System.Reflection.Assembly]::LoadWithPartialName("System.Drawing")
$provider = New-Object System.Security.Cryptography.SHA1CryptoServiceProvider

foreach ($location in $args)
{
    $files=get-childitem $location | where{$_.Extension -match "jpg|jpeg"}
    foreach ($f in $files)
        {
        $bitmap = New-Object -TypeName System.Drawing.Bitmap -ArgumentList $f.FullName
        $stream = New-Object -TypeName System.IO.MemoryStream
        $bitmap.Save($stream)

        $hashbytes = $provider.ComputeHash($stream.ToArray())
        $hashstring = ""
        foreach ($byte in $hashbytes) 
            {$hashstring += $byte.tostring("x2")}  
        $f.FullName
        $hashstring
        echo ""
        }
} 

This produces the same hash regardless of the input file, so something is still not quite correct.

Answer 3

This is a PowerShell V2.0 advanced function implemention. It is a bit long but I have verified it gives the same hashcode (generated from the bitmap pixels) on the same picture but with different metadata and file sizes. This is a pipeline capable version that also accepts wildcards and literal paths:

function Get-BitmapHashCode
{
    [CmdletBinding(DefaultParameterSetName="Path")]
    param(
        [Parameter(Mandatory=$true, 
                   Position=0, 
                   ParameterSetName="Path", 
                   ValueFromPipeline=$true, 
                   ValueFromPipelineByPropertyName=$true,
                   HelpMessage="Path to bitmap file")]
        [ValidateNotNullOrEmpty()]
        [string[]]
        $Path,

        [Alias("PSPath")]
        [Parameter(Mandatory=$true, 
                   Position=0, 
                   ParameterSetName="LiteralPath", 
                   ValueFromPipelineByPropertyName=$true,
                   HelpMessage="Path to bitmap file")]
        [ValidateNotNullOrEmpty()]
        [string[]]
        $LiteralPath
    )

    Begin {
        Add-Type -AssemblyName System.Drawing
        $sha = new-object System.Security.Cryptography.SHA256Managed
    }

    Process {
        if ($psCmdlet.ParameterSetName -eq "Path")
        {
            # In -Path case we may need to resolve a wildcarded path
            $resolvedPaths = @($Path | Resolve-Path | Convert-Path)
        }
        else 
        {
            # Must be -LiteralPath
            $resolvedPaths = @($LiteralPath | Convert-Path)
        }

        # Find PInvoke info for each specified path       
        foreach ($rpath in $resolvedPaths) 
        {           
            Write-Verbose "Processing $rpath"
            try {
                $bmp    = new-object System.Drawing.Bitmap $rpath
                $stream = new-object System.IO.MemoryStream
                $writer = new-object System.IO.BinaryWriter $stream
                for ($w = 0; $w -lt $bmp.Width; $w++) {
                    for ($h = 0; $h -lt $bmp.Height; $h++) {
                        $pixel = $bmp.GetPixel($w,$h)
                        $writer.Write($pixel.ToArgb())
                    }
                }
                $writer.Flush()
                [void]$stream.Seek(0,'Begin')
                $hash = $sha.ComputeHash($stream)
                [BitConverter]::ToString($hash) -replace '-',''
            }
            finally {
                if ($bmp)    { $bmp.Dispose() }
                if ($writer) { $writer.Close() }
            }
        }  
    }
}

Answer 4

Here's a powershell script that produces an SHA256 hash on only the bytes of the image as extracted using LockBits. This should produce a unique hash for each file that is different. Please note, that I didn't include the file iterating code, however it should be a relatively simple task to replace the currently hardcode c:\test.bmp with a foreach directory iterator. The variable $final contains the hex - ascii string of the final hash.

[System.Reflection.Assembly]::LoadWithPartialName("System.Drawing")
[System.Reflection.Assembly]::LoadWithPartialName("System.Drawing.Imaging")
[System.Reflection.Assembly]::LoadWithPartialName("System.Security")


$bmp = [System.Drawing.Bitmap]::FromFile("c:\\test.bmp")
$rect = [System.Drawing.Rectangle]::FromLTRB(0, 0, $bmp.width, $bmp.height)
$lockmode = [System.Drawing.Imaging.ImageLockMode]::ReadOnly               
$bmpData = $bmp.LockBits($rect, $lockmode, $bmp.PixelFormat);
$dataPointer = $bmpData.Scan0;
$totalBytes = $bmpData.Stride * $bmp.Height;
$values = New-Object byte[] $totalBytes
[System.Runtime.InteropServices.Marshal]::Copy($dataPointer, $values, 0, $totalBytes);                
$bmp.UnlockBits($bmpData);

$sha = new-object System.Security.Cryptography.SHA256Managed
$hash = $sha.ComputeHash($values);
$final = [System.BitConverter]::ToString($hash).Replace("-", "");

Perhaps the equivalent C# code will also aid you in understanding:

private static String ImageDataHash(FileInfo imgFile)
{
    using (Bitmap bmp = (Bitmap)Bitmap.FromFile(imgFile.FullName))
    {                
        BitmapData bmpData = bmp.LockBits(new Rectangle(0, 0, bmp.Width, bmp.Height), System.Drawing.Imaging.ImageLockMode.ReadOnly, bmp.PixelFormat);
        IntPtr dataPointer = bmpData.Scan0;
        int totalBytes = bmpData.Stride * bmp.Height;
        byte[] values = new byte[totalBytes];                
        System.Runtime.InteropServices.Marshal.Copy(dataPointer, values, 0, totalBytes);                
        bmp.UnlockBits(bmpData);
        SHA256 sha = new SHA256Managed();
        byte[] hash = sha.ComputeHash(values);
        return BitConverter.ToString(hash).Replace("-", "");                
    }
}