- Notifications
You must be signed in to change notification settings - Fork0
The official motion detection engine of ShinobiCCTV by@kevinGodell. This fork should be seen as a release manager. For updated code for this module please visit the following link.
License
ShinobiCCTV/EyeOfHorus
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
Measure differences between pixel arrays extracted from pam images. Works well with node modulepipe2pam to extract pam images from an ffmpeg pipe. Supportedtupltypes arergb,rgb_alpha,grayscale, andblackandwhite. It is currently being used for a video motion detection project.
npm install pam-diff --save
To run the example below, also install pipe2pam:
npm install pipe2pam --save
The followingexample uses ffmpeg to connect to a rtsp ip camera video feed and generates 1000 downscaled rgb24 pam images at a rate of 1 per second. The pam images are piped from ffmpeg's stdout into pipe2pam to parse them into into pam objects. The pam objects are then piped into pam-diff to measure pixel differences. For each compared pixel that has adifference that exceeds the setting, it will be calculated to determine the percent of difference. If thepercent of changed pixels exceeds the setting, adiff event will be emitted which contains a data object containing details. This example also shows how to take the pam image that triggered the diff event and convert it to a jpeg using ffmpeg.
const P2P = require('pipe2pam');const PamDiff = require('pam-diff');const ChildProcess = require('child_process');const spawn = ChildProcess.spawn;const execFile = ChildProcess.execFile;const params = [ '-loglevel', 'quiet', /* use hardware acceleration */ '-hwaccel', 'auto', //vda, videotoolbox, none, auto /* use an artificial video input */ /*'-re', '-f', 'lavfi', '-i', 'testsrc=size=1920x1080:rate=15',*/ /* use an rtsp ip cam video input */ '-rtsp_transport', 'tcp', '-i', 'rtsp://192.168.1.4:554/user=admin_password=pass_channel=1_stream=0.sdp', /* set output flags */ '-an', '-c:v', 'pam', '-pix_fmt', 'rgb24',//rgba, rgb24, gray '-f', 'image2pipe', '-vf', 'fps=1,scale=320:180',//1920:1080 scaled down: 400:225, 384:216, 368:207, 352:198, 336:189, 320:180 //'fps=1,scale=iw*1/6:ih*1/6', '-frames', '1000', 'pipe:1'];const ffmpeg = spawn('ffmpeg', params);console.log(ffmpeg.spawnargs.join(' '));ffmpeg.on('error', (error) => { console.log(error);});ffmpeg.on('exit', (code, signal) => { console.log('exit', code, signal);});const p2p = new P2P();let counter = 0;p2p.on('pam', (data) => { //you do not have to listen to this event if you are just piping this data to pam-diff console.log(`received pam ${++counter}`);});const pamDiff = new PamDiff({grayscale: 'average', difference: 5, percent: 5});pamDiff.on('diff', (data) => { console.log(data); //comment out the following line if you want to use ffmpeg to create a jpeg from the pam image that triggered an image difference event if(true){return;} const date = new Date(); let name = `${date.getUTCFullYear()}-${date.getUTCMonth() + 1}-${date.getUTCDate()}_${date.getHours()}-${date.getUTCMinutes()}-${date.getUTCSeconds()}-${date.getUTCMilliseconds()}`; for (const region of data.trigger) { name += `(${region.name}=${region.percent})`; } const jpeg = `${name}.jpeg`; const ff = execFile('ffmpeg', ['-f', 'pam_pipe', '-c:v', 'pam', '-i', 'pipe:0', '-c:v', 'mjpeg', '-pix_fmt', 'yuvj422p', '-q:v', '1', '-huffman', 'optimal', jpeg]); ff.stdin.end(data.pam); ff.on('exit', (data) => { if (data === 0) { console.log(`FFMPEG clean exit after creating ${jpeg}`); } else { throw new Error('FFMPEG is not working with current parameters'); } });});ffmpeg.stdout.pipe(p2p).pipe(pamDiff);
About
The official motion detection engine of ShinobiCCTV by@kevinGodell. This fork should be seen as a release manager. For updated code for this module please visit the following link.
Topics
Resources
License
Stars
Watchers
Forks
Releases
Packages0
Languages
- JavaScript100.0%