Converting cv2 images to PySDL2 surfaces for blitt

问题:

I'm using the new PySDL2 package, trying to interface it with my existing OpenCV code. I want to take an image captured from a webcam via the cv2 python interface to OpenCV and use PySDL2 to show it in a window on screen. I think I figured out how to convert the cv2 image format to a PySDL2 surface properly, but at the end of the code below, all I get is a black window. Any pointers on where I have gone awry would be greatly appreciated!

#grab a frame from a webcam
import cv2
vc = cv2.VideoCapture(0)
junk,image = vc.read()

#convert image to sdl format (?)
import sdl2
sbuf = image.tostring()
simage = sdl2.SDL_CreateRGBSurfaceFrom(sbuf,image.shape[0],image.shape[1],24,3*image.shape[0],sdl2.SDL_PIXELFORMAT_BGRA8888,0xff0000, 0x00ff00, 0x0000ff, 0)

#create a window
sdl2.SDL_Init(sdl2.SDL_INIT_VIDEO)
windowSize = (640,480)
window = sdl2.SDL_CreateWindow(b"Hello World",sdl2.SDL_WINDOWPOS_CENTERED, sdl2.SDL_WINDOWPOS_CENTERED,windowSize[0], windowSize[1], sdl2.SDL_WINDOW_SHOWN)
windowSurface = sdl2.SDL_GetWindowSurface(window)

#try to blit the sdl-formatted image to the window
sdl2.SDL_BlitSurface(simage,None,windowSurface,None)
sdl2.SDL_UpdateWindowSurface(window)
sdl2.SDL_FreeSurface(simage)

# pump events to get the window to show and update
while True:
    sdl2.SDL_PumpEvents()

回答1:

Solved it!

#import necessary modules
import cv2
import sdl2
import sdl2.ext
import numpy

windowSize = (640,480)

#initialize the camera
vc = cv2.VideoCapture(0)
vc.set(cv2.cv.CV_CAP_PROP_FRAME_WIDTH, windowSize[0])
vc.set(cv2.cv.CV_CAP_PROP_FRAME_HEIGHT, windowSize[1])

#grab and show a first frame from the camera
junk,image = vc.read()
cv2.imshow('0',image)

#initialize sdl2
sdl2.ext.init()
window = sdl2.ext.Window("test", size=windowSize)
window.show()
windowSurf = sdl2.SDL_GetWindowSurface(window.window)
windowArray = sdl2.ext.pixels3d(windowSurf.contents)

while True: #keep reading to have a live feed from the cam
    junk,image = vc.read()
    image = numpy.insert(image,3,255,axis=2) #add alpha
    image = numpy.rot90(image) #rotate dims
    numpy.copyto(windowArray, image)
    window.refresh()

I'm not sure why showing a first frame from the camera using cv2.imshow is necessary, but without that part, the sdl2 window never appears.

回答2:

i do not have enough reputation to add comment, though its been long time ago, hope this can help someone,

you can add a flag for the window creation

window = sdl2.ext.Window("test", size=windowSize, flag=sdl2.SDL_WINDOW_SHOWN)