We Keep Coding

Remove remainders from bitmap after background removal

I have this image and I want to remove the background to isolate the green picture. The background is not completely black but it contains some pixels having the same color of other pixels inside the green picture.
I have used this
private void ButtonFilterClick(object sender, EventArgs e)
{
PixelFormat pxf = PixelFormat.Format24bppRgb;
Bitmap bitmap = ((Bitmap)(_smartLabForm.pictureBox1.Image));
Rectangle rect = new Rectangle(0, 0, bitmap.Width, bitmap.Height);
BitmapData bmpData =
bitmap.LockBits(rect, ImageLockMode.ReadWrite, pxf);
IntPtr ptr = bmpData.Scan0;
int numBytes = bmpData.Stride * bitmap.Height;
byte[] rgbValues = new byte[numBytes];
Marshal.Copy(ptr, rgbValues, 0, numBytes);
for (int counter = 0; counter < rgbValues.Length; counter += 3)
{
if (rgbValues[counter] < 15 &&
rgbValues[counter + 1] < 15 &&
rgbValues[counter + 2] < 15)
{
rgbValues[counter] = 255;
rgbValues[counter + 1] = 255;
rgbValues[counter + 2] = 255;
}
}
Marshal.Copy(rgbValues, 0, ptr, numBytes);
bitmap.UnlockBits(bmpData);
_smartLabForm.Refresh();
}
and what I obtain is this:
How can I remove the "noise" remaining without damage the green picture and without affecting the performance?
Thank you?

This is actually a quite complex topic in computer vision (image segmentation). Covering the advanced techniques would be way too broad. But here is quick and simple idea that may get the job done:
Increase the threshold enough that all background pixels fall below it. When checking if a pixel should be removed, also compare all pixels in a certain neighborhood (e.g. circular radius) with the threshold. Only remove it if they are all below the threshold.
That way you remove pixels less agressively when you are near the feature region.

C# - RGB Buffer from Bitmap different from C++

I'm using a third party DLL which has as parameter a RGB buffer.
I have used the following code to read RGB buffer from Bitmap:
private byte[] GetBGRValues(Bitmap bmp)
{
// Lock the bitmap's bits.
Rectangle rect = new Rectangle(0, 0, bmp.Width, bmp.Height);
System.Drawing.Imaging.BitmapData bmpData = bmp.LockBits(rect, System.Drawing.Imaging.ImageLockMode.ReadOnly, PixelFormat.Format24bppRgb);
// Get the address of the first line.
IntPtr ptr = bmpData.Scan0;
// Declare an array to hold the bytes of the bitmap.
int bytes = Math.Abs(bmpData.Stride) * bmp.Height;
byte[] rgbValues = new byte[bytes];
// Copy the RGB values into the array.
System.Runtime.InteropServices.Marshal.Copy(ptr, rgbValues, 0, bytes);
bmp.UnlockBits(bmpData);
return rgbValues;
}
The problem is that the generated RGB buffer is not correct. If I open this buffer in IrfanView, supplying correct parameters, the generated image is not correct (looks like it is shifted).
If a get a buffer that I read using C++ code it works.
I have noticed that bmpData.Stride is 1 unity greater than what I was expecting (width * channels). (I know that .NET uses 4 bytes alignment).
The question is: why is the RGB buffer not correct?

You noticed right - you need to take Stride into account. In general you cannot simply copy image in one Copy call. Stride include both row length and padding and could be greater then row length. So you need to copy only bytes you need from each row, ignore padding bytes and advance to next row by adding Stride.
I guess this is what you see with your code:
- original image and expected result
- invalid result without stride
Here is working code:
public static byte[] GetBGRValues(Bitmap bmp)
{
var rect = new Rectangle(0, 0, bmp.Width, bmp.Height);
var bmpData = bmp.LockBits(rect, System.Drawing.Imaging.ImageLockMode.ReadOnly, bmp.PixelFormat);
var rowBytes = bmpData.Width * Image.GetPixelFormatSize(bmp.PixelFormat) / 8;
var imgBytes = bmp.Height * rowBytes;
byte[] rgbValues = new byte[imgBytes];
var ptr = bmpData.Scan0;
for (var i = 0; i < bmp.Height; i++)
{
Marshal.Copy(ptr, rgbValues, i * rowBytes, rowBytes);
ptr += bmpData.Stride; // next row
}
bmp.UnlockBits(bmpData);
return rgbValues;
}
More details you can read in this answer: Byte Array to Image Conversion
Also maybe this image will help you to understand Stride purpose:
You need to skip white area at the right when you getting bytes from Bitmap.

Be sure you watch that the order is B-G-R instead of R-G-B
You can try this unsafe code which converts the values to uint. So you have RGB converted to uint
/// <summary>
/// Locks a Bitmap into system memory.
/// </summary>
public unsafe void LockBits()
{
var width = this.Bitmap.Width;
var height = this.Bitmap.Height;
var imageLockMode = ImageLockMode.UserInputBuffer;
// Setting imageLockMode
imageLockMode = imageLockMode | ImageLockMode.ReadOnly;
imageLockMode = imageLockMode | ImageLockMode.WriteOnly;
// Save the bouunds
this._bounds = new Rectangle(0, 0, width, height);
// Create Pointer
var someBuffer = new uint[width*height];
// Pin someBuffer
fixed (uint* buffer = someBuffer) //pin
{
// Create new bitmap data.
var temporaryData = new BitmapData
{
Width = width,
Height = height,
PixelFormat = PixelFormat.Format32bppArgb,
Stride = width*4,
Scan0 = (IntPtr) buffer
};
// Get the data
this.BitmapData = this.Bitmap.LockBits(this._bounds, imageLockMode, PixelFormat.Format32bppArgb,
temporaryData);
// Set values
this.Buffer = someBuffer;
}
}

I remember a library I working on years ago - the colors were shifted strangely. The underlying Microsoft Library had a (feature) in that the RGB had been reversed inside the library - unbeknownst to us we tried a transform and discovered that little gem.
Also do not forget the Alpha Channel in your C# buffer.
Bitmap color channels in memory are represented in the order Blue, Green, Red and Alpha despite being commonly referred to by abbreviation ARGB!
http://softwarebydefault.com/2013/03/22/bitmap-swap-argb/

Image Manipulation c#

I have the following code which takes an array of bytes which i generated and writes them out to this bitmap. If i set the pixel format to Format4bppIndexed, then i get a readable image repeating width wise 4 times, if i set it to Format1bppIndexed(which is the correct setting) then i get one big unreadable image.
The image was a decoded Jbig2 image , i know the bytes are correct i can't seem to figure out how to get it into a 1bpp readable format.
Does anyone have any advice on that matter
Bitmap bitmap = new Bitmap(width, height, PixelFormat.Format1bppIndexed);
//Create a BitmapData and Lock all pixels to be written
BitmapData bmpData = bitmap.LockBits(
new Rectangle(0, 0, bitmap.Width, bitmap.Height),
ImageLockMode.WriteOnly, bitmap.PixelFormat);
//Copy the data from the byte array into BitmapData.Scan0
Marshal.Copy(newarray, 0, bmpData.Scan0, newarray.Length);
//Unlock the pixels
bitmap.UnlockBits(bmpData);

The following may work although, if I remember correctly, Stride sometimes has an effect and a simple block-copy won't suffice (line by line must be used instead).
Bitmap bitmap = new Bitmap(
width,
height,
System.Drawing.PixelFormat.Format16bppGrayScale
);
To handle the Stride you'd want:
BitmapData^ data = bitmap->LockBits(oSize,
ImageLockMode::ReadOnly, bitmap->PixelFormat);
try {
unsigned char *pData = (unsigned char *)data->Scan0.ToPointer();
for( int x = 0; x < bmpImage->Width; ++x )
{
for( int y = 0; y < bmpImage->Height; ++y )
{
// Note: Stride is data width of scan line rounded up
// to 4 byte boundary.
// Requires use of Stride, not (width * pixelWidth)
int ps = y*bmpImage->Width*(nBitsPerPixel / 8)
+ x * (nBitsPerPixel / 8);
int p = y * data->Stride + x * (nBitsPerPixel / 8);
Byte lo = newarray[ps + 1];
Byte hi = newarray[ps + 0];
pData[p + 1] = lo;
pData[p + 0] = hi;
}
}
} finally {
bmpImage->UnlockBits(data);
}
Note: This was written in C++/CLI. Let me know if you need C# equivalents for any of the operations here. (Also, I pulled it from a read from bitmap rather than a write to bitmap so it may yet be a bit rough, but should hopefully give you the idea...)

I figured this out Although i'm still not sure why it should matter.
Based on this stackoverflow posting How can I load the raw data of a 48bpp image into a Bitmap?
I used the WPF classes instead of the GDI and wrote the code like this
var bitmap = new WriteableBitmap(width, height, 96, 96, System.Windows.Media.PixelFormats.BlackWhite, null);
bitmap.WritePixels(new System.Windows.Int32Rect(0, 0, width, height), newarray, stride, 0);
MemoryStream stream3 = new MemoryStream();
var encoder = new TiffBitmapEncoder ();
encoder.Frames.Add(BitmapFrame.Create(bitmap));
encoder.Save(stream3);
This correctly creates the image.
If anyone has any insight into why this might be the case please comment below
The port which now mostly works(lots of cleanup code) was based on a java implementation of JPedal Big2 Decoder to .NET. If anyone knows anyone interested send them here
https://github.com/devteamexpress/JBig2Decoder.NET

C# - Marshal.Copy : Attempted to read or write protected memory

"Attempted to read or write protected memory. This is often an indication that other memory is corrupt."
I'm experiencing this error in the Marshal.Copy portion of my code. I do believe that my data is not corrupted nor protected.
I was wondering in what case does this occur.
I have a List<> of bitmaps. This only occurs when I process the first index [0].
So here's how I did it :
- First, I used this code [This code gets the pixel data of a bitmap] :
Bitmap tmp_bitmap = BitmapFromFile[0];
Rectangle rect = new Rectangle(0, 0, tmp_bitmap.Width, tmp_bitmap.Height);
System.Drawing.Imaging.BitmapData bmpData =
tmp_bitmap.LockBits(rect, System.Drawing.Imaging.ImageLockMode.ReadWrite,
PixelFormat.Format24bppRgb);
int length = bmpData.Stride * bmpData.Height;
byte[] bytes = new byte[length];
// Copy bitmap to byte[]
Marshal.Copy(bmpData.Scan0, bytes, 0, length);
tmp_bitmap.UnlockBits(bmpData);
It works fine, no errors occur.
Then, I apply this code [ This will remove the pixel data line scan padding ]:
byte[] bytes = new byte[bmpData.Width * bmpData.Height * 3];
for (int y = 0; y < bmpData.Height; ++y) {
IntPtr mem = (IntPtr)((long)bmpData.Scan0 + y * bmpData.Stride * 3);
Marshal.Copy(mem, bytes, y * bmpData.Width * 3, bmpData.Width * 3); //This is where the exception is pointed.
}
It gives me that error whenever I'm processing the first image -- second to last, no problem at all.
I hope you can help me with this.
Thank you in advance.

You seem to be considering 3 times the stride for every row; your code will only work for the first third of the image; after that you have indeed gone outside of your allowed range. Basically:
bmpData.Scan0 + y * bmpData.Stride * 3
looks really dodgy. The "stride" is the number of bytes (including padding) used by every line. Typically, that would be just:
bmpData.Scan0 + y * bmpData.Stride

Why are pixel values of an image different between reading methods?

This puzzles me for last two hours. Reading an image file results in different pixel values between imread in Matlab and Image.FromFile in C#?
aa=imread('myfile.tif')
max(aa(:)) = 248 in matlab
In C#
var image2Array = imageToByteArray((Bitmap) Image.FromFile("myfile.tif"));
byte maxx = 0;
foreach(var a in image2Array)
{
maxx = Math.Max(maxx, a);
}
//maxx = 255
Futhermore, in Matlab,
aa(1,1) = 13,
aa(1,2) = 13
but in C#
image2Array[0]=17,
image2Array[1]=0
They should be the same.
BTW, in this case, pixel type is uint8. so there is no dimensional difference.
If you ask me how I got byte array from Image, I used MSDN document to make this method.
public byte[] imageToByteArray(Bitmap bmp)
{
// Lock the bitmap's bits.
Rectangle rect = new Rectangle(0, 0, bmp.Width, bmp.Height);
BitmapData bmpData = bmp.LockBits(
rect,
ImageLockMode.ReadWrite,
bmp.PixelFormat);
// Get the address of the first line.
IntPtr ptr = bmpData.Scan0;
// Declare an array to hold the bytes of the bitmap.
int bytes = Math.Abs(bmpData.Stride)*bmp.Height;
byte[] rgbValues = new byte[bytes];
// Copy the RGB values into the array.
System.Runtime.InteropServices.Marshal.Copy(ptr, rgbValues, 0, bytes);
// Unlock the bits.
bmp.UnlockBits(bmpData);
return rgbValues;
}
What did I do wrong here? I suspect that they use different reading algorithms because two resulting images look same.
UPDATE:
I don't think there is anything wrong with what I was doing. I concluded that reading tif as a bitmap was the cause of the problem. To confirm this theory,
I displayed the two images and they looked exactly the same. So there is no mistake on my part, I think.
I tried to read the same file with opencv and its pixel values were exactly the same as the ones from matlab. This was surprisingly to me. I would very cautiously use Bitmap in C# from now on.

Your imageToByteArray method does return a byte array, but you can't assume each byte is a pixel. The PixelFormat determines how the pixel data is stored in the byte array.
The best site I've seen that documents this is Bob Powell's lockbits page.
If the PixelFormat is Format8bppIndexed, then this (untested) code should give you the color values for each pixel.
var bmp = (Bitmap)Bitmap.FromFile("myfile.tif");
// ******* Begin copying your imageToByteArray method
// Lock the bitmap's bits.
Rectangle rect = new Rectangle(0, 0, bmp.Width, bmp.Height);
BitmapData bmpData = bmp.LockBits(
rect,
ImageLockMode.ReadWrite,
bmp.PixelFormat);
// Get the address of the first line.
IntPtr ptr = bmpData.Scan0;
// Declare an array to hold the bytes of the bitmap.
int bytes = Math.Abs(bmpData.Stride) * bmp.Height;
byte[] imageData = new byte[bytes];
// Copy the RGB values into the array.
System.Runtime.InteropServices.Marshal.Copy(ptr, imageData, 0, bytes);
// Unlock the bits.
bmp.UnlockBits(bmpData);
// ******* End copying your imageToByteArray method
// Now loop through each pixel... The byte array contains extra bytes
// used for padding so we can't just loop through every byte in the array.
// This is done by using the Stride property on bmpData.
for (int y = 0; y < bmpData.Height; y++)
{
for (int x = 0; x < bmpData.Width; x++)
{
var offset = (y * bmpData.Stride) + x;
// The byte in the image array gives the offset into the palette
var paletteIndex = imageData[offset];
// Given the offset, find the matching color in the palette
var color = bmp.Palette.Entries[offset];
// Look at the color value here...
}
}

TIFF has many formats, you are attempting to read it as a bitmap.
I suggest reading it using a proprietary TIFF reader instead : Good Tiff library for .NET