32位CBitmap转灰度图问题求助：现有代码实现异常

Alright, let's work through your 32-bit CBitmap grayscale problem. Your code snippet cuts off early, but I can spot common pitfalls that usually trip up this kind of implementation, plus share a complete, tested version of the function.

Common Issues in Partial Code #

First, let's call out red flags from what you shared:

• You're using ::GetDC(HWND_DESKTOP) but not releasing it—this will leak a DC handle over time. Always pair GetDC with ReleaseDC, or better yet, use memory DCs for bitmap operations.
• Your BITMAPINFO setup is incomplete: 32-bit bitmaps require explicit settings for biBitCount, biCompression, and biSizeImage to avoid undefined behavior.

Complete Working Implementation #

Here's a revised version of your function that addresses these issues and properly converts a 32-bit bitmap to grayscale while retaining the original background/alpha channel:

void CMeetingScheduleAssistantApp::SetBitmapAsGrayScale(HBITMAP hbmp) { 
    // Early exit if invalid bitmap
    if (!hbmp) return; 

    // Retrieve basic bitmap properties
    BITMAP bm;
    ::GetObject(hbmp, sizeof(bm), &bm);

    // Only process 32-bit bitmaps (adjust if you need support for other formats)
    if (bm.bmBitsPixel != 32) {
        return;
    }

    // Initialize BITMAPINFO for DIBSection (direct pixel access)
    BITMAPINFO bi = {0};
    bi.bmiHeader.biSize = sizeof(BITMAPINFOHEADER);
    bi.bmiHeader.biWidth = bm.bmWidth;
    bi.bmiHeader.biHeight = bm.bmHeight;
    bi.bmiHeader.biPlanes = 1;
    bi.bmiHeader.biBitCount = 32;
    bi.bmiHeader.biCompression = BI_RGB;
    bi.bmiHeader.biSizeImage = bm.bmWidth * bm.bmHeight * 4; // 4 bytes per 32-bit pixel

    // Create a DIBSection to access pixel data directly
    void* pBits = nullptr;
    HBITMAP hDib = ::CreateDIBSection(nullptr, &bi, DIB_RGB_COLORS, &pBits, nullptr, 0);
    if (!hDib || !pBits) {
        return;
    }

    // Set up memory DCs to copy original bitmap data to DIBSection
    HDC hdcMemSrc = ::CreateCompatibleDC(nullptr);
    HDC hdcMemDst = ::CreateCompatibleDC(nullptr);
    HBITMAP hOldSrc = (HBITMAP)::SelectObject(hdcMemSrc, hbmp);
    HBITMAP hOldDst = (HBITMAP)::SelectObject(hdcMemDst, hDib);

    // Copy original bitmap to DIBSection
    ::BitBlt(hdcMemDst, 0, 0, bm.bmWidth, bm.bmHeight, hdcMemSrc, 0, 0, SRCCOPY);

    // Convert each pixel to grayscale
    DWORD* pPixel = static_cast<DWORD*>(pBits);
    int totalPixels = bm.bmWidth * bm.bmHeight;

    for (int i = 0; i < totalPixels; ++i) {
        // Extract B, G, R, and Alpha components (32-bit DIB uses BGRA format)
        BYTE blue = (pPixel[i] >> 0) & 0xFF;
        BYTE green = (pPixel[i] >> 8) & 0xFF;
        BYTE red = (pPixel[i] >> 16) & 0xFF;
        BYTE alpha = (pPixel[i] >> 24) & 0xFF;

        // Calculate grayscale using standard luminance weights (for natural-looking grayscale)
        BYTE grayValue = static_cast<BYTE>(0.299 * red + 0.587 * green + 0.114 * blue);

        // Reconstruct pixel: keep original alpha, set B/G/R to grayscale value
        pPixel[i] = (alpha << 24) | (grayValue << 16) | (grayValue << 8) | grayValue;
    }

    // Copy the grayscale data back to the original bitmap
    ::BitBlt(hdcMemSrc, 0, 0, bm.bmWidth, bm.bmHeight, hdcMemDst, 0, 0, SRCCOPY);

    // Clean up all resources to avoid leaks
    ::SelectObject(hdcMemSrc, hOldSrc);
    ::SelectObject(hdcMemDst, hOldDst);
    ::DeleteDC(hdcMemSrc);
    ::DeleteDC(hdcMemDst);
    ::DeleteObject(hDib);
}

Key Notes About This Implementation #

• Resource Safety: Every created DC and bitmap is properly released with DeleteDC/DeleteObject, and we restore the original objects in each DC with SelectObject.
• Direct Pixel Access: Using CreateDIBSection lets us manipulate pixel data directly, which is more reliable than GetBitmapBits/SetBitmapBits for 32-bit formats.
• Natural Grayscale: The luminance formula (0.299*R + 0.587*G + 0.114*B) is the industry standard for converting RGB to grayscale—it mimics how human eyes perceive color brightness.
• Alpha Channel Preservation: The code retains the original alpha (transparency) channel. If you don't need transparency, you can set the alpha component to 0xFF (fully opaque) instead.

内容的提问来源于stack exchange，提问作者Andrew Truckle