bytedeco/javacv

I'm passing a Frame to a BufferedImage to get a Array of each RGB channel - is there a more direct path?

Open

#2,097 opened on Sep 14, 2023

View on GitHub
 (1 comment) (0 reactions) (0 assignees)Java (1,583 forks)batch import
help wantedquestion

Repository metrics

Stars
 (6,985 stars)
PR merge metrics
 (No merged PRs in 30d)

Description

I feel like I'm converting when I don't need to be, and that there is likely a much more direct path.

Input: A short lived org.bytedeco.javacv.Frame read from a FFmpegFrameGrabber.

Output: the image fully decoded and normalized in a very basic format of:

class DecodedImage(
    internal val width: Int,
    internal val height: Int,
    private val red: IntArray = IntArray(width * height),
    private val green: IntArray = IntArray(width * height),
    private val blue: IntArray = IntArray(width * height),
)

Ints per channel per pixel are a lot, but I need them for some later summing.

Right now I'm using a Java2DFrameConverter to convert the Frame to a BufferedImage, then reading out pixels from the BufferedImage -> Raster -> dataBuffer.

That seems... excessive. Is there a more direct way of blowing out a Frame to a fully flattened DecodedImage? I poked around in the Java2DFrameConverter, and it handles a lot of image formats.

Contributor guide