Index out of range exception while reading pixel data

Question

Related post: Stride of the BitmapData is different than the original dimension.

I have taken the source code from here and modified it.

The code is generating a variety of exceptions in different occasions.

.

Error in BitmapLocker.cs

At the following line in Lock(),

// Copy data from IntegerPointer to _imageData
Marshal.Copy(IntegerPointer, _imageData, 0, _imageData.Length);

The following exception is being generated:

An unhandled exception of type 'System.AccessViolationException' occurred in mscorlib.dll

Additional information: Attempted to read or write protected memory. This is often an indication that other memory is corrupt.

For the following driver code,

        double[,] mask = new double[,] 
                                    {   
                                    { .11, .11, .11, }, 
                                    { .11, .11, .11, }, 
                                    { .11, .11, .11, }, 
                                    };
        Bitmap bitmap = ImageDataConverter.ToBitmap(mask);

        BitmapLocker locker = new BitmapLocker(bitmap);

        locker.Lock();

        for (int i = 0; i < bitmap.Width; i++)
        {
            for (int j = 0; j < bitmap.Height; j++)
            {
                Color c = locker.GetPixel(i, j);

                locker.SetPixel(i, j, c);
            }
        }

        locker.Unlock();

At the following line in GetPixel(),

        if (i > dataLength)
        {
            throw new IndexOutOfRangeException();
        }

An unhandled exception of type 'System.IndexOutOfRangeException' occurred in Simple.ImageProcessing.Framework.dll

Additional information: Index was outside the bounds of the array.

.

At the following line in SetPixel(),

if (ColorDepth == 8)
{
    _imageData[i] = color.B;
}

An unhandled exception of type 'System.Exception' occurred in Simple.ImageProcessing.Framework.dll

Additional information: (0, 0), 262144, Index was outside the bounds of the array., i=262144

.

Error in Driver program

At the line,

Color c = bmp.GetPixel(i, j);

An unhandled exception of type 'System.InvalidOperationException' occurred in System.Drawing.dll

Additional information: Bitmap region is already locked.

---

---

Source Code:

public class BitmapLocker : IDisposable
{
    //private properties
    Bitmap _bitmap = null;
    bool _isLocked = false;
    BitmapData _bitmapData = null;
    private byte[] _imageData = null;

    //public properties
    public IntPtr IntegerPointer { get; private set; }
    public int Width { get { return _bitmap.Width; } }
    public int Height { get { return _bitmap.Height; } }
    public int Stride { get { return _bitmapData.Stride; } }
    public int ColorDepth { get { return Bitmap.GetPixelFormatSize(_bitmap.PixelFormat); } }
    public int Channels { get { return ColorDepth / 8; } }
    public int PaddingOffset { get { return _bitmapData.Stride - (_bitmap.Width * Channels); } }
    public PixelFormat ImagePixelFormat { get { return _bitmap.PixelFormat; } }
    public bool IsGrayscale { get { return Grayscale.IsGrayscale(_bitmap); } }

    //Constructor
    public BitmapLocker(Bitmap source)
    {
        IntegerPointer = IntPtr.Zero;
        this._bitmap = source;
    }

    /// Lock bitmap
    public void Lock()
    {
        if (_isLocked == false)
        {
            try
            {
                // Lock bitmap (so that no movement of data by .NET framework) and return bitmap data
                _bitmapData = _bitmap.LockBits(
                                                new Rectangle(0, 0, _bitmap.Width, _bitmap.Height),
                                                ImageLockMode.ReadWrite,
                                                _bitmap.PixelFormat);

                // Create byte array to copy pixel values
                int noOfBitsNeededForStorage = _bitmapData.Stride * _bitmapData.Height;

                int noOfBytesNeededForStorage = noOfBitsNeededForStorage / 8;

                _imageData = new byte[noOfBytesNeededForStorage * ColorDepth];//# of bytes needed for storage

                IntegerPointer = _bitmapData.Scan0;

                // Copy data from IntegerPointer to _imageData
                Marshal.Copy(IntegerPointer, _imageData, 0, _imageData.Length);

                _isLocked = true;
            }
            catch (Exception)
            {
                throw;
            }
        }
        else
        {
            throw new Exception("Bitmap is already locked.");
        }
    }

    /// Unlock bitmap
    public void Unlock()
    {
        if (_isLocked == true)
        {
            try
            {
                // Copy data from _imageData to IntegerPointer
                Marshal.Copy(_imageData, 0, IntegerPointer, _imageData.Length);

                // Unlock bitmap data
                _bitmap.UnlockBits(_bitmapData);

                _isLocked = false;
            }
            catch (Exception)
            {
                throw;
            }
        }
        else
        {
            throw new Exception("Bitmap is not locked.");
        }
    }

    public Color GetPixel(int x, int y)
    {
        Color clr = Color.Empty;

        // Get color components count
        int channels = ColorDepth / 8;

        // Get start index of the specified pixel
        int i = (Height - y - 1) * Stride + x * channels;

        int dataLength = _imageData.Length - channels;

        if (i > dataLength)
        {
            throw new IndexOutOfRangeException();
        }

        if (ColorDepth == 32) // For 32 bpp get Red, Green, Blue and Alpha
        {
            byte b = _imageData[i];
            byte g = _imageData[i + 1];
            byte r = _imageData[i + 2];
            byte a = _imageData[i + 3]; // a
            clr = Color.FromArgb(a, r, g, b);
        }
        if (ColorDepth == 24) // For 24 bpp get Red, Green and Blue
        {
            byte b = _imageData[i];
            byte g = _imageData[i + 1];
            byte r = _imageData[i + 2];
            clr = Color.FromArgb(r, g, b);
        }
        if (ColorDepth == 8)
        // For 8 bpp get color value (Red, Green and Blue values are the same)
        {
            byte c = _imageData[i];
            clr = Color.FromArgb(c, c, c);
        }
        return clr;
    }

    public void SetPixel(int x, int y, Color color)
    {

            // Get color components count
            int cCount = ColorDepth / 8;

            // Get start index of the specified pixel
            int i = ((Height - y -1) * Stride + x * cCount);                

            try
            {
            if (ColorDepth == 32) // For 32 bpp set Red, Green, Blue and Alpha
            {
                _imageData[i] = color.B;
                _imageData[i + 1] = color.G;
                _imageData[i + 2] = color.R;
                _imageData[i + 3] = color.A;
            }
            if (ColorDepth == 24) // For 24 bpp set Red, Green and Blue
            {
                _imageData[i] = color.B;
                _imageData[i + 1] = color.G;
                _imageData[i + 2] = color.R;
            }
            if (ColorDepth == 8)
            // For 8 bpp set color value (Red, Green and Blue values are the same)
            {
                _imageData[i] = color.B;
            }
        }
        catch(Exception ex)
        {
            throw new Exception("("+x+", "+y+"), "+_imageData.Length+", "+ ex.Message+", i=" + i);
        }
    }

    public void Dispose()
    {
        Dispose(true);
        GC.SuppressFinalize(this);
    }

    protected virtual void Dispose(bool disposing)
    {
        if (disposing)
        {
            // free managed resources
            _bitmap = null;
            _bitmapData = null;_imageData = null;IntegerPointer = IntPtr.Zero;
        }
        // free native resources if there are any.

        //private properties



        //public properties

    }
}

.

ImageDataConverter.cs

    public static Bitmap ToBitmap(double[,] input)
    {
        int width = input.GetLength(0);
        int height = input.GetLength(1);

        Bitmap output = Grayscale.CreateGrayscaleImage(width, height);

        BitmapData data = output.LockBits(new Rectangle(0, 0, width, height),
                                            ImageLockMode.WriteOnly,
                                            output.PixelFormat);

        int pixelSize = System.Drawing.Image.GetPixelFormatSize(PixelFormat.Format8bppIndexed) / 8;

        int offset = data.Stride - width * pixelSize;

        double Min = 0.0;
        double Max = 255.0;

        unsafe
        {
            byte* address = (byte*)data.Scan0.ToPointer();

            for (int y = 0; y < height; y++)
            {
                for (int x = 0; x < width; x++)
                {
                    double v = 255 * (input[x, y] - Min) / (Max - Min);

                    byte value = unchecked((byte)v);

                    for (int c = 0; c < pixelSize; c++, address++)
                    {
                        *address = value;
                    }
                }

                address += offset;
            }
        }

        output.UnlockBits(data);

        return output;
    }

Here is the picture I used for the test,

Answer 1

int noOfBitsNeededForStorage = _bitmapData.Stride * _bitmapData.Height;

That is the most essential bug in the code. Stride * Height are the number of bytes needed for storage. So it doesn't make the _imageData array large enough and IndexOutOfRangeException is the expected outcome.

int pixelSize = System.Drawing.Image.GetPixelFormatSize(PixelFormat.Format8bppIndexed) / 8;

Lots of possible mishaps from this statement. It hard-codes the pixel format to 8bpp but that is not the actual pixel format that the LockBits() call used. Which was output.PixelFormat. Notably fatal on the sample image, although it is not clear how it is used in the code, 8bpp is a very awkward pixel format since it requires a palette. The PNG codec will create a 32bpp image in memory, even though the original file uses 8bpp. You must use output.PixelFormat here to get a match with the locked data and adjust the pixel writing code accordingly. Not clear why it is being used at all, the SetPixel() method provided by the library code should already be good enough.

int dataLength = _imageData.Length - channels;

Unclear what that statement tries to do, subtracting the number of channels is not a sensible operation. It will generate a spurious IndexOutOfRangeException. There is no obvious reason to help, the CLR already provides array index checking on the _imageData array. So just delete that code.

Additional information: Bitmap region is already locked.

Exception handling in the code is not confident, a possible reason for this exception. In general it must be noted that the underlying bitmap is completely inaccessible, other than through _imageData, after the Lock() method was called and Unlock() wasn't called yet. Best way to do this is with try/finally with the Unlock() call in the finally block so you can always be sure that the bitmap doesn't remain locked by accident.

byte c = _imageData[i];

This is not correct, except in the corner case of an 8bpp image that has a palette that was explicitly created to handle grayscale images. The default palette for an 8bpp image does not qualify that requirement nor is it something you can blindly rely on when loading images from a file. Indexed pixel formats where a dreadful hack that was necessary in the early 1990s because video adapters where not yet powerful enough. It no longer makes any sense at all today. Note that SetPixel() also doesn't handle a 16-bit pixel formats. And that the PNG codec will never create an 8bpp memory image and cannot encode an 8bpp file. Best advice is to eliminate 8bpp support completely to arrive at more reliable code.

In fact, the point of directly accessing pixel data is to make image manipulation fast. There is only one pixel format that consistently produces fast code, it is Format32bppArgb. The pixels can now be accessed with an int* instead of a byte*, moving pixels ~4 times faster. And no special tweaks are necessary to deal with stride or special-case the code for methods like SetPixel(). So pass that format into LockBits(), the codec will do the work necessary if the actual image format is not 32bpp as well.

I should note that Format32bppPArgb is the fast pixel format for displaying images on the screen since it is compatible with the pixel format used by all modern video adapters. But that isn't the point of this code and dealing with the pre-multiplied alpha is awkward.