Question

I am trying to get the pixel colour from an image displayed by the webcam. I want to see how the pixel colour is changing with time.

My current solution sucks a LOT of CPU, it works and gives me the correct answer, but I am not 100% sure if I am doing this correctly or I could cut some steps out.

- (IBAction)addFrame:(id)sender
{
    // Get the most recent frame
    // This must be done in a @synchronized block because the delegate method that sets the most recent frame is not called on the main thread
    CVImageBufferRef imageBuffer;

    @synchronized (self) {
        imageBuffer = CVBufferRetain(mCurrentImageBuffer);
    }

    if (imageBuffer) {
        // Create an NSImage and add it to the movie
        // I think I can remove some steps here, but not sure where.
        NSCIImageRep *imageRep = [NSCIImageRep imageRepWithCIImage:[CIImage imageWithCVImageBuffer:imageBuffer]];
        NSSize n = {320,160 };
        //NSImage *image = [[[NSImage alloc] initWithSize:[imageRep size]] autorelease];
        NSImage *image = [[[NSImage alloc] initWithSize:n] autorelease];
        [image addRepresentation:imageRep];
        CVBufferRelease(imageBuffer);


        NSBitmapImageRep* raw_img = [NSBitmapImageRep imageRepWithData:[image TIFFRepresentation]];
        NSLog(@"image width is %f", [image size].width);
        NSColor* color = [raw_img colorAtX:1279 y:120];

        float colourValue = [color greenComponent]+ [color redComponent]+ [color blueComponent];
        [graphView setXY:10 andY:200*colourValue/3];

    NSLog(@"%0.3f", colourValue);

Any help is appreciated and I am happy to try other ideas.
Thanks guys.

Answer 1

There are a couple of ways that this could be made more efficient. Take a look at the imageFromSampleBuffer: method in this Tech Q&A, which presents a cleaner way of getting from a CVImageBufferRef to an image (the sample uses a UIImage, but it’s practically identical for an NSImage).

You can also pull the pixel values straight out of the CVImageBufferRef without any conversion. Once you have the base address of the buffer, you an calculate the offset of any pixel and just read the values from there.