Raspberry Camera API

From Rsewiki
(Difference between revisions)
Jump to: navigation, search
(Raspicam)
(Format check)
 
(23 intermediate revisions by one user not shown)
Line 1: Line 1:
 
Back to [[Robobot]]
 
Back to [[Robobot]]
  
===Raspicam===
+
== Camera use ==
  
To be used from C++ direct or through openCV.
+
Raspberry pi operating system version "bullseye" uses a new camera library.
  
A rather short raspberry pi camera API is available from https://www.uco.es/investiga/grupos/ava/node/40.
+
The info is taken from https://www.raspberrypi.com/documentation/accessories/camera.html#libcamera-and-libcamera-apps?ref=https://githubhelp.com
  
And can be downloaded from https://sourceforge.net/projects/raspicam/files/,
+
=== LibCamera ===
  
or get the file from sourceforge with no GUI, with:
+
If not already installed, install with
  
  cd ~/Downloads
+
  sudo apt install libcamera-apps
wget --no-check-certificate -O raspicam-0.1.9.zip https://downloads.sourceforge.net/project/raspicam/raspicam-0.1.9.zip?r=https%3A%2F%2Fsourceforge.net%2Fprojects%2Fraspicam%2F%3Fsource%3Dtyp_redirect&ts=1486483484&use_mirror=netix
+
  
Once fetched and available on the raspberry, then
+
It should not be needed to install libcamera-dev
Unpack and install:
+
unzip raspicam-0.1.9.zip
+
cd raspicam-0.1.9
+
mkdir build
+
cd build
+
cmake ..
+
make -j3
+
sudo make install
+
sudo ldconfig
+
  
The library is installed in /usr/local, and therefore to make cmake find it the path needs to be added to startup configuration, in file ~/.bashrc
+
==== Camera streaming ====
  
cd
+
The libcamera library can do streaming out of the box with the command:
nano ~/.bashrc
+
add at the end:
+
export CMAKE_PREFIX_PATH=/usr/local/lib
+
source ~/.bashrc  # to implement export
+
  
Test with openCV
+
  libcamera-vid -n -t 0 --inline --listen --rotation 0 --framerate 30 --width 1920 --height 1080 -o tcp://0.0.0.0:8888
  raspicam_cv_test
+
That takes 100 images and saves 3.
+
  
=== Camera streamer ===
+
It will listen to port 8888 for TCP connections on the localhost.
  
To download the project you will need a source control system called git. It may not be installed on a fresh image. I know it’s not on the lite image. So you may need to install it.
+
From another PC the stream can be displayed by VLC, (or the same parameter in the VLC "media" -> "open network stream")
  sudo apt-get install git
+
   
 +
vlc tcp/h264://192.168.2.14:8888
  
Now that you have git installed, use it to clone a copy of the mjpg-streamer to your Pi.
+
if 192.168.2.14 is the IP of the Raspberry pi.
 +
The streaming will terminate when the VLC stops the connection.
  
git clone https://github.com/jacksonliam/mjpg-streamer.git
+
A start-script is placed in the "local" home folder.
  
After the cloned copy of the mjpeg-stream has been coppied to the raspberry-pi, follow these steps.
+
./stream_cam.sh
  
cd mjpg-streamer/
+
with examples of different solutions.
cd mjpg-streamer-experimental/
+
sudo apt-get install cmake
+
# sudo apt-get install python-imaging
+
# sudo apt-get install libjpeg-dev
+
make CMAKE_BUILD_TYPE=Debug
+
sudo make install
+
  
Make a script to start the streamer
+
== camera from OpenCV ==
  
cd
+
The raspberry cameras seem to prefer 10 bit/pixel in Bayer RGGB format - other formats are not consistent for both new and old camera.
nano start-stream.sh
+
  
copy this into the file (uncomment only one line - the others are examples for other resolutions)
+
=== Format check ===
  
#!/bin/bash
+
Install package v4l-utils (also installed by default)
#mjpg_streamer -i "input_raspicam.so -y 1600 -x 1200 -fps 5 -rot 90" -o output_http.so
+
mjpg_streamer -i "input_raspicam.so -y 972 -x 1296 -fps 3 -rot=0" -o output_http.so
+
#mjpg_streamer -i "input_raspicam.so -y 600 -x 800 -fps 3 -rot=0" -o output_http.so
+
#mjpg_streamer -i "input_raspicam.so -y 640 -x 480 -fps 5" -o output_http.so
+
  
Change the file to be executable
+
sudo apt install v4l-utils
  
  chmod +x start-stream.sh
+
The supported image formats from both cameras are:
 +
 
 +
  v4l2-ctl -d/dev/video0 --list-formats
 +
 
 +
ioctl: VIDIOC_ENUM_FMT
 +
        Type: Video Capture
 +
        [0]: 'BA81' (8-bit Bayer BGBG/GRGR)
 +
        [1]: 'pBAA' (10-bit Bayer BGBG/GRGR Packed)
 +
        [2]: 'BG10' (10-bit Bayer BGBG/GRGR)
 +
 
 +
For the used format 'BG10' the frame sizes supported is found
 +
 
 +
v4l2-ctl -d /dev/video0 --list-framesizes BG10
 +
 
 +
ioctl: VIDIOC_ENUM_FRAMESIZES
 +
        Size: Discrete 2592x1944
 +
        Size: Discrete 1920x1080
 +
        Size: Discrete 1296x972
 +
        Size: Discrete 640x480
 +
 
 +
 
 +
 
 +
But formats can be converted to other formats, visible as other video devices e.g /dev/video12.
 +
 
 +
v4l2-ctl -d/dev/video12 --list-formats
 +
 
 +
ioctl: VIDIOC_ENUM_FMT
 +
        Type: Video Capture Multiplanar
 +
        [0]: 'YUYV' (YUYV 4:2:2)
 +
        [1]: 'YVYU' (YVYU 4:2:2)
 +
        [2]: 'VYUY' (VYUY 4:2:2)
 +
        [3]: 'UYVY' (UYVY 4:2:2)
 +
        [4]: 'YU12' (Planar YUV 4:2:0)
 +
        [5]: 'YV12' (Planar YVU 4:2:0)
 +
        [6]: 'RGB3' (24-bit RGB 8-8-8)
 +
        [7]: 'BGR3' (24-bit BGR 8-8-8)
 +
        [8]: 'BGR4' (32-bit BGRA/X 8-8-8-8)
 +
        [9]: 'RGBP' (16-bit RGB 5-6-5)
 +
        [10]: 'NV12' (Y/CbCr 4:2:0)
 +
        [11]: 'NV21' (Y/CrCb 4:2:0)
 +
 
 +
Format BA81 did not give a proper result for both cameras.
 +
I didn't succeed in using the converted formats, maybe because they are based on the BA81 format.
 +
Format 'BG10' works for both cameras and is easily converted to RGB format using OpenCV (https://docs.opencv.org/3.4/db/d64/tutorial_js_colorspaces.html).

Latest revision as of 14:49, 4 January 2022

Back to Robobot

Contents

[edit] Camera use

Raspberry pi operating system version "bullseye" uses a new camera library.

The info is taken from https://www.raspberrypi.com/documentation/accessories/camera.html#libcamera-and-libcamera-apps?ref=https://githubhelp.com

[edit] LibCamera

If not already installed, install with

sudo apt install libcamera-apps

It should not be needed to install libcamera-dev

[edit] Camera streaming

The libcamera library can do streaming out of the box with the command:

libcamera-vid -n -t 0 --inline --listen --rotation 0 --framerate 30 --width 1920 --height 1080 -o tcp://0.0.0.0:8888

It will listen to port 8888 for TCP connections on the localhost.

From another PC the stream can be displayed by VLC, (or the same parameter in the VLC "media" -> "open network stream")

vlc tcp/h264://192.168.2.14:8888

if 192.168.2.14 is the IP of the Raspberry pi. The streaming will terminate when the VLC stops the connection.

A start-script is placed in the "local" home folder.

./stream_cam.sh

with examples of different solutions.

[edit] camera from OpenCV

The raspberry cameras seem to prefer 10 bit/pixel in Bayer RGGB format - other formats are not consistent for both new and old camera.

[edit] Format check

Install package v4l-utils (also installed by default)

sudo apt install v4l-utils

The supported image formats from both cameras are:

v4l2-ctl -d/dev/video0 --list-formats
ioctl: VIDIOC_ENUM_FMT
       Type: Video Capture
       [0]: 'BA81' (8-bit Bayer BGBG/GRGR)
       [1]: 'pBAA' (10-bit Bayer BGBG/GRGR Packed)
       [2]: 'BG10' (10-bit Bayer BGBG/GRGR)

For the used format 'BG10' the frame sizes supported is found

v4l2-ctl -d /dev/video0 --list-framesizes BG10
ioctl: VIDIOC_ENUM_FRAMESIZES
       Size: Discrete 2592x1944
       Size: Discrete 1920x1080
       Size: Discrete 1296x972
       Size: Discrete 640x480


But formats can be converted to other formats, visible as other video devices e.g /dev/video12.

v4l2-ctl -d/dev/video12 --list-formats
ioctl: VIDIOC_ENUM_FMT
       Type: Video Capture Multiplanar
       [0]: 'YUYV' (YUYV 4:2:2)
       [1]: 'YVYU' (YVYU 4:2:2)
       [2]: 'VYUY' (VYUY 4:2:2)
       [3]: 'UYVY' (UYVY 4:2:2)
       [4]: 'YU12' (Planar YUV 4:2:0)
       [5]: 'YV12' (Planar YVU 4:2:0)
       [6]: 'RGB3' (24-bit RGB 8-8-8)
       [7]: 'BGR3' (24-bit BGR 8-8-8)
       [8]: 'BGR4' (32-bit BGRA/X 8-8-8-8)
       [9]: 'RGBP' (16-bit RGB 5-6-5)
       [10]: 'NV12' (Y/CbCr 4:2:0)
       [11]: 'NV21' (Y/CrCb 4:2:0)

Format BA81 did not give a proper result for both cameras. I didn't succeed in using the converted formats, maybe because they are based on the BA81 format. Format 'BG10' works for both cameras and is easily converted to RGB format using OpenCV (https://docs.opencv.org/3.4/db/d64/tutorial_js_colorspaces.html).

Personal tools
Namespaces

Variants
Actions
Navigation
Toolbox