Look into Bitmap images

What's a Bitmap image?

I'm not going to explain the differences between raster and vector images, nor the difference between bitmap image with extension .bmp and jpg/png images. Herein, the bitmap refers to raster images with .bmp extension only.

Bitmap is uncompressed, and the image is represented with arrays, namely, RGB arrays.

What's the storage of a bitmap

File structure of a bitmap

  • The bitmaps stored on the physical devices contains serval parts:
  1. BITMAPINFO
    1. BITMAPINFOHEADER
    2. RGBQUAD
    2. The stored data in array

The definition are

typedef struct tagBITMAPFILEHEADER {
    WORD  bfType;		// 2 bytes
    DWORD bfSize;		// 4 bytes
    WORD  bfReserved1;	// 2 bytes
    WORD  bfReserved2;  // 2 bytes
    DWORD bfOffBits;    // 4 bytes
} BITMAPFILEHEADER, *LPBITMAPFILEHEADER, *PBITMAPFILEHEADER;
typedef struct tagBITMAPINFO {
    BITMAPINFOHEADER bmiHeader;
    RGBQUAD          bmiColors[1];		// the '1' here is pretty tricky
} BITMAPINFO, *LPBITMAPINFO, *PBITMAPINFO;
typedef struct tagBITMAPINFOHEADER {
    DWORD biSize;			 // 8 bytes
    LONG  biWidth;			// 4 bytes
    LONG  biHeight;			// 4 bytes
    WORD  biPlanes;			//
    WORD  biBitCount;
    DWORD biCompression;
    DWORD biSizeImage;
    LONG  biXPelsPerMeter;
    LONG  biYPelsPerMeter;
    DWORD biClrUsed;
    DWORD biClrImportant;
} BITMAPINFOHEADER, *PBITMAPINFOHEADER;
typedef struct tagRGBQUAD {
    BYTE rgbBlue;
    BYTE rgbGreen;
    BYTE rgbRed;
    BYTE rgbReserved;
} RGBQUAD;

Let's use Lena.bmp as an illustration:


(Well, this blog doesn;t support BMP image, please see the attach file)

Open it with a hex editor, such as WinHex, and it shows

The highlighted 14 bytes are BITMAPFILEHEADER,

WORD bfType -> 0x4D42
DWORD bfSize -> 0x00010438 = 66616
WORD bfReserved1 -> 0x0000 = 0
WORD bfReserved2 -> 0x0000 = 0
DWORD bfOffBits -> 0x00000436 = 1078

  • bfType shows that this file is a bitmap file;
  • bfSize is the file size, as we can see below, the file sis 66,614 bytes;
  • bfReserved1 and bfReserved2 must be zeros
  • bfOffBits is the offset, in bytes, from the beginning of the BITMAPFILEHEADER structure to the bitmap bits.

e:\temp\code\MFC_PICTURE_TEST\Look into Bitmap images.assets\1566902617332.png

And this is the start of array data,

seen what I mean?

The next is BITMAPINFOHEADER, and the data are highlighted in different colors,

and these values can be calculated as

DWORD biSize = 0x00000028 = 40
LONG biWidth = 0x00000100 = 256
LONG biHeight = 0x00000100 = 256
WORD biPlanes = 0x0001 = 1
WORD biBitCount = 0x0008 = 8
DWORD biCompression = 0x00000000 = 0
DWORD biSizeImage = 0x00010002 = 65,538
LONG biXPelsPerMeter = 0x00000B12 = 2,834
LONG biYPelsPerMeter = 0x00000B12 = 2,834
DWORD biClrUsed = 0x000000 = 0
DWORD biClrImportant = 0x000000 = 0

And we can see,

0x0436 + 0x00010002 = 0x00010438

1,078 + 65,538 = 66,616

i.e.

data_start_offset + data_length = file_end


It's the start of data, however, since Lena.bmp is a gray image, all the three channels are same, you can find a pure red/blue image, you will find that, the bitmap is stored in [B, G, R, reserved] order, therefore, if you want to interpret in RGB mode, you should be careful with it.

[C++]Read Bitmap File

The following code are mainly from this website, slightly modified.

typedef unsigned int Uint8;
Uint8* datBuff[2] = { nullptr, nullptr }; // Header buffers
Uint8* pixels = nullptr; // Pixels
BITMAPFILEHEADER* bmpHeader = nullptr; // Header
BITMAPINFOHEADER* bmpInfo = nullptr; // Info 

// The file... We open it with it's constructor
std::ifstream file(_T("Lena.bmp"), std::ios::binary);
if (!file){
std::cout << "Failure to open bitmap file.\n";
return 1;
}
// Allocate byte memory that will hold the two headers
datBuff[0] = new Uint8[sizeof(BITMAPFILEHEADER)];
datBuff[1] = new Uint8[sizeof(BITMAPINFOHEADER)];

file.read((char*)datBuff[0], sizeof(BITMAPFILEHEADER));
file.read((char*)datBuff[1], sizeof(BITMAPINFOHEADER));
// Construct the values from the buffers
bmpHeader = (BITMAPFILEHEADER*)datBuff[0];
bmpInfo = (BITMAPINFOHEADER*)datBuff[1];
// Check if the file is an actual BMP file
if (bmpHeader->bfType != 0x4D42){
std::cout << "File isn't a bitmap file\n";
return 2;
}
// First allocate pixel memory
pixels = new Uint8[bmpInfo->biSizeImage];

// Go to where image data starts, then read in image data
file.seekg(bmpHeader->bfOffBits);
file.read((char*)pixels, bmpInfo->biSizeImage);
// First allocate pixel memory
pixels = new Uint8[bmpInfo->biSizeImage];

// Go to where image data starts, then read in image data
file.seekg(bmpHeader->bfOffBits);
file.read((char*)pixels, bmpInfo->biSizeImage);

// Set width and height to the values loaded from the file
int w = bmpInfo->biWidth;
int h = bmpInfo->biHeight;

// We're almost done. We have our image loaded, however it's not in the right format.
// .bmp files store image data in the BGR format, and we have to convert it to RGB.
// Since we have the value in bytes, this shouldn't be to hard to accomplish
Uint8 tmpRGB = 0; // Swap buffer
for (unsigned long i = 0; i < bmpInfo->biSizeImage; i += 3){
    tmpRGB        = pixels[i];
    pixels[i]     = pixels[i + 2];
    pixels[i + 2] = tmpRGB;
}

delete[] datBuff[0];
delete[] datBuff[1];
delete[] pixels;

Create Bitmap Image in memory, and draw on screen

The code is mainly from this webpage, slightly modified.

Again, the reference are given:

  1. BITMAPINFO
  2. BITMAPINFOHEADER
  3. RGBQUAD
  4. SetDIBitsToDevice
void CChildView::OnPaint() 
{
	CPaintDC dc(this); // device context for painting
	HDC hdc = dc.m_hDC;

	int nWidth = 100;
	int nHeight = 100;
	int nChannels = 3;
	int nImages = 3;	// three images in a row, presented in a column
	auto nSize = nWidth * nHeight * nChannels * nImages;
	unsigned char* data = new unsigned char[nSize] {0};
	//memset(data, 0xFF, nSize);


	long split = nWidth * nHeight / 4,
		nPixel = nWidth * nHeight;		// Pixels in an image
	// Fill the different images for differnt channels
	unsigned char* dat = data;
	for (int i = 0; i < nPixel; ++i, dat += 3)		// Image 0
		dat[0] = (i / nWidth) * (255.0 / nHeight);
	for (int i = 0; i < nPixel; ++i, dat += 3)		// Image 1
		dat[1] = (i / nWidth) * (255.0 / nHeight);
	for (int i = 0; i < nPixel; ++i, dat += 3)		// Image 2
		dat[2] = (i / nWidth) * (255.0 / nHeight);

	

	// Allocate enough memory for the BITMAPINFOHEADER and 256 RGBQUAD palette entries
	// NOTE: the pallet bytes are ONLY NEEDED for color LUT images ...
	// not needed here
	int nColors = 0;		// In BI_RGB image, the look-up-table isn't used, so the size can be zero
	LPBITMAPINFO lpbi = (LPBITMAPINFO) new BYTE[sizeof(BITMAPINFOHEADER)+ (nColors * sizeof(RGBQUAD))];
	
	// These are all the members of the bitmap header struct
	lpbi->bmiHeader.biSize   = sizeof(BITMAPINFOHEADER); // bytes
	lpbi->bmiHeader.biWidth  = nWidth;                   // pixels units
	lpbi->bmiHeader.biHeight = -nHeight * 3;             // negative =  = top down; pos =
	// origin LLeft
	lpbi->bmiHeader.biPlanes = 1;    // must be 1
	lpbi->bmiHeader.biBitCount = 24; // can be 32 for 4-byte pixels
	// (Upper byte ignored)
	lpbi->bmiHeader.biCompression = BI_RGB; // BI_RGB means uncompressed
	lpbi->bmiHeader.biSizeImage = 0;        // size of img in bytes; 0 okay
	// for BI_RGB
	lpbi->bmiHeader.biXPelsPerMeter = 0;	// physical device information unavailable
	lpbi->bmiHeader.biYPelsPerMeter = 0;    // physical device information unavailable
	lpbi->bmiHeader.biClrUsed       = 0;	// Look-Up-Tables (LUT) only
	lpbi->bmiHeader.biClrImportant  = 0;	// LUTs only

	// Draw the image into the CRT device
	::SetDIBitsToDevice(
		hdc,                 // handle to DC
		0, 0,                // x-y-coord of destination upper-left corner
		nWidth, nHeight * 3, // width-height of source rectangle
			                 // three images present in one column
		0, 0,                // x-y-coord of source upper-left corner
		0,                   // uStartScan,                                // first scan line in array
		nHeight * 3,         // number of scan lines ...
		data,                // array of DIB bits
		lpbi,                // bitmap information
		DIB_RGB_COLORS);     // RGB vs. palette indexes ... RGB means raw
	delete[] data;
	delete[] lpbi;
}

As we can see from the result, the array is stored in [B, G, R] sequence, and no more reserved byte is need.

Besides, if you uncomment the memset(data, 0xFF, nSize); at the 12th line, which means you set all the other values to be 255, you'll get a more beautiful image as

If you are a Chinese, this blog may help you.

The source code can be found in: 链接: https://pan.baidu.com/s/12NCP1tQjtLOdDccODOwpEg 提取码: 4uka 复制这段内容后打开百度网盘手机App,操作更方便哦

posted @ 2019-08-27 20:04  Troy_Daniel  阅读(344)  评论(0编辑  收藏  举报