如何在Android MLKit中将图形叠加层转换为Bitmap

Got it, let's walk through how to convert that segmentation overlay graphic into a Bitmap. Based on the SegmentationGraphic class logic from the ML Kit sample, here's a straightforward, actionable approach:

Core Idea #

Instead of drawing the segmentation mask directly to a preview canvas (like the sample does), we'll create a mutable Bitmap, attach a Canvas to it, and run the same drawing logic on this Bitmap-linked canvas.

Step 1: Prepare the Bitmap & Canvas #

First, create a Bitmap that matches the dimensions of your input image or camera frame. Then get a Canvas instance tied to this Bitmap—this is where we'll draw the segmentation overlay.

// Replace these with your actual input image/frame dimensions
int inputWidth = ...; 
int inputHeight = ...;

// Create a mutable Bitmap to hold the segmentation overlay
Bitmap segmentationBitmap = Bitmap.createBitmap(inputWidth, inputHeight, Bitmap.Config.ARGB_8888);
Canvas bitmapCanvas = new Canvas(segmentationBitmap);

Step 2: Adapt the Segmentation Drawing Logic #

Take the core mask-rendering code from the sample's draw(Canvas) method and adapt it to draw on your bitmap's canvas. Here's the key part:

// Get your ML Kit segmentation result (adjust this line to match your result retrieval)
SegmentationResult segmentationResult = ...;
SegmentationMask mask = segmentationResult.getSegmentationMask();
ByteBuffer maskBuffer = mask.getBuffer();
int maskWidth = mask.getWidth();
int maskHeight = mask.getHeight();

// Set up the paint for the overlay (semi-transparent green as in the sample)
Paint overlayPaint = new Paint();
overlayPaint.setColor(Color.argb(128, 0, 255, 0));

// Scale the canvas to match the mask to input image size
bitmapCanvas.save();
bitmapCanvas.scale(
    (float) inputWidth / maskWidth,
    (float) inputHeight / maskHeight
);

// Iterate through the mask buffer and draw foreground regions
for (int y = 0; y < maskHeight; y++) {
    for (int x = 0; x < maskWidth; x++) {
        float foregroundConfidence = maskBuffer.getFloat();
        // Only draw pixels where foreground confidence exceeds your threshold
        if (foregroundConfidence > 0.5) {
            bitmapCanvas.drawRect(x, y, x + 1, y + 1, overlayPaint);
        }
    }
}
bitmapCanvas.restore();

Step 3: Use the Resulting Bitmap #

Now segmentationBitmap holds your segmentation overlay as a standalone Bitmap. You can display it in an ImageView, save it to storage, or combine it with the original image as needed.

Quick Notes #

• Adjust the threshold (0.5 in the example) and overlay color to fit your use case.
• If your input image has rotation or scaling transformations (like camera preview orientation), add those to the bitmapCanvas before drawing (mirroring the transformations used in the sample's GraphicOverlay).

内容的提问来源于stack exchange，提问作者Hemant Bharti