A simple program demonstrating use of the Raspicam with OpenCV and a Frame Buffer to a SPI TFT LCD This is a program I put together from other works mainly as a demo. I wanted to get a Raspberry Pi Zero to use the Rasperry Pi Noir Camera working with the Pi Zero but I also wanted to put the image out to a SPI TFT 240x320 LCD that I had on hand.
Some important things to note about this program: a lot depends on what the camera brings in. I used a Noir and this may differ in the color that comes in, so your setting Camera.set( CV_CAP_PROP_FORMAT, CV_8UC1 ); is apt to be critical here. But I had for a long time some very funky colors on the LCD, looking like an acid scene from a 1960s movie. So the LCD itself and it's own color capabilities also wreak havoc here. This line: cvtColor(frame, frame2, COLOR_GRAY2BGR565); was found to work to finally give out what is essentially a gray image. I think other applications, like a color camera and more capable LCD will differ in the settings. Note also if you are new to frame buffer usage, be mindful of the following:
- the capture dimensions of the camera. You can almost get a "focus/expand" effect narrowing the FOV of the camera if you mix this up.
- the detected resolution of your screen versus the virtual resolution. You can make the virtal bigger.
- the segfaults that will occur when you are putting "too many bits" into the frame buffer
- the "color conver" of OpenCV, what it got from the camera, and the needs of where the image is going. HDMI screens notably did not need color conversion and the buffer gleaned from the camera could be sent straight to output. Expect to experiment.
To compile and link the sample program:
g++ -o raspistr raspistr.cpp pkg-config opencv --cflags --libs -lraspicam -lraspicam_cv -L/path/to/dir_containing/libpng -lpng
yes it's rather redundant to use Raspicam and OpenCV...
Also when you have hooked up your SPI LCD screen, enter this command:
sudo modprobe fbtft_device name=tm022hdh26 rotate=90
Some other helpful links that helped me in this project, and some of the code for this program was cobled from samples found around the web.
https://neverbenever.wordpress.com/2015/05/18/how-to-use-pi-camera-with-opencv-for-raspberry-pi/ http://marcosgildavid.blogspot.com/2014/02/getting-ili9341-spi-screen-working-on.html http://raspberrycompote.blogspot.com/2016/ https://github.com/notro/fbtft/wiki