Java Motion Detection using webcam
I'm in a group that is trying to detect motion from people walking through a space and then for each separate person, to display a fish in an "aquarium". Right now we're using Processing and OpenCV to do facial detection and then for each face, we create a fish.
The problem is that the system runs at a really low frame rate. We need to get the frame rate up by a large factor, which is why we're trying to figure out how to use motion detection.
Our teacher suggested that we use threading in processing and check less often for people, but we're really confused by how to do the threading.
Any suggestions about how we should handle this would be really really appreciated. Thank you.
Here is the code:
import hypermedia.video.*; // Imports the OpenCV library
import java.awt.Rectangle;
import ddf.minim.analysis.*;
import ddf.minim.*;
OpenCV opencv; // Creates a new OpenCV object
Minim minim;
AudioInput in;
FFT fft;
int w;
PImage fade;
PImage movementImg; // Creates a new PImage to hold the movement image
ArrayList bubbles; // Creates an ArrayList to hold the Bubble objects
PImage bubblePNG; // Creates a PImage that will hold the image of the bubble
int randPos;
PImage fishImg;
PImage fishImg2;
PImage fishImg3;
PImage fishImg4;
PImage fishImg5;
PImage sharkImg;
PImage clockImg;
PImage backImg;
int sharkX=480;
int sharkY=height/2;
int sharkMoves = 480;
int sharkSpeed=40;
int flagForShark=0;
int flagForNotification=0;
ArrayList psystems;
int NotificationX = 10;
int NotificationY = 10;
//clock
int cx, cy;
float secondsRadius;
float minutesRadius;
float hoursRadius;
float clockDiameter;
void setup(){
size ( 640, 480 ); // Window size of 640 x 480
opencv = new OpenCV( this ); // Initialises the OpenCV library
opencv.capture( 640, 480 ); // Sets the capture size to 640 x 480
opencv.cascade( OpenCV.CASCADE_FRONTALFACE_ALT ); //// load the FRONTALFACE description file
movementImg = new PImage(640, 480 ); // Initialises the PImage that holds the movement image
bubbles = new ArrayList(); // Initialises the ArrayList
bubblePNG = loadImage("bubble.png"); // Load the bubble image into memory
smooth();
fishImg = loadImage("purpleFish.png");
fishImg2 = loadImage("fish2.png");
fishImg3 = loadImage("fish3.png");
fishImg4 = loadImage("fish4.png");
fishImg5 = loadImage("fish5.png");
sharkImg = loadImage("shark.png");
clockImg = loadImage("clock.png");
backImg = loadImage("bg01.png");
fill(61,36,9);
int radius = min(100, 100) / 2;
secondsRadius = radius * 0.72;
minutesRadius = radius * 0.60;
hoursRadius = radius * 0.50;
clockDiameter = radius * 1.8;
cx = 50;
cy = 50;
//Sound stuff
minim = new Minim(this);
in = minim.getLineIn(Minim.STEREO, 512);
fft = new FFT(in.bufferSize(),in.sampleRate());
fft.logAverages(60,7);
}
void youareloud(){
fft.forward(in.mix);
for(int i=0; i<fft.avgSize();i++){
if(fft.getAvg(i) > 3){
randPos = 160*(int)random(0, 5);
bubbles.add(new Bubble( randPos+(int)random(-10, 10), 480, (int)random(10,25), (int)random(10,25))); // Adds a new bubble to the array with a random x position
}
}
for ( int i = 0; i < bubbles.size(); i++ ){ // For every bubble in the bubbles array
Bubble _bubble = (Bubble) bubbles.get(i); // Copies the current bubble into a temporary object
if(_bubble.update() == 1){ // If the bubble's update function returns '1'
bubbles.remove(i); // then remove the bubble from the array
_bubble = null; // and make the temporary bubble object null
i--; // since we've removed a bubble from the array, we need to subtract 1 from i, or we'll skip the next bubble
}else{ // If the bubble's update function doesn't return '1'
bubbles.set(i, _bubble); // Copys the updated temporary bubble object back into the array
_bubble = null; // Makes the temporary bubble object null.
}
}
}
void draw(){
opencv.read(); // Captures a frame from the camera
opencv.flip(OpenCV.FLIP_HORIZONTAL); // Flips the image horizontally
// background(loadImage("data/underwater_640x480_stretched.jpg"));//drwa detected environemtn
background(backImg);
faces();
youareloud();
extras();
}
class Bubble{
int bubbleX, bubbleY, bubbleWidth, bubbleHeight; //Some variables to hold information about the bubble
int randSize = (int)random(10, 20);
Bubble ( int bX, int bY, int bW, int bH ){ //The class constructor- sets the values when a new bubble object is made
bubbleX = bX;
bubbleY = bY;
bubbleWidth = bW;
bubbleHeight = bH;
}
int update(){ //The Bubble update function
int movementAmount; //Create and set a variable to hold the amount of white pixels detected in the area where the bubble is
movementAmount = 0;
for( int y = bubbleY; y < (bubbleY + (bubbleHeight-1)); y++ ){
//For loop that cycles through all of the pixels in the area the bubble occupies
for( int x = bubbleX; x < (bubbleX + (bubbleWidth-1)); x++ ){
if ( x < width && x > 0 && y < height && y > 0 ){
//If the current pixel is within the screen bondaries
if (brightness(movementImg.pixels[x + (y * width)]) > 127){
//and if the brightness is above 127 (in this case, if it is white)
movementAmount++;
//Add 1 to the movementAmount variable.
}
}
}
}
if (movementAmount > 5){ // If more than 5 pixels of movement are detected in the bubble area
//poppedBubbles++; // Add 1 to the variable that holds the number of popped bubbles
return 1; // Return 1 so that the bubble object is destroyed
} else { // If less than 5 pixels of movement are detected,
//bubbleY += 10; // increase the y position of the bubble so that it falls down
bubbleY -= 10; // increase the y position of the bubble so that it falls down
if (bubbleY < 0){ // If the bubble has dropped off of the bottom of the screen
return 1; // Return '1' so that the bubble object is destroyed
}
image(bubblePNG, bubbleX, bubbleY,randSize,randSize); // Draws the bubble to the screen
return 0; // Returns '0' so that the bubble isn't destroyed
}
}
}
void faces(){
Rectangle[] faces = opencv.detect();
noFill();
stroke(255,0,0);
opencv.absDiff(); // Creates a difference image
opencv.convert(OpenCV.GRAY); // Converts to greyscale
opencv.blur(OpenCV.BLUR, 3); // Blur to remove camera noise
opencv.threshold(20); // Thresholds to convert to black and white
movementImg = opencv.image(); // Puts the OpenCV buffer into an image object
opencv.remember(OpenCV.SOURCE, OpenCV.FLIP_HORIZONTAL); // Remembers the camera image so we can generate a difference image next frame. Since we've
for( int i=0; i<faces.length; i++ ) {
//image( opencv.image(), faces[i].x, faces[i].y, faces[i].width, faces[i].height ); // display the image in memory on the right
// opencv.loadImage( "/Users/sang/Desktop/home.png", ); // load image from file
// opencv.convert( GRAY );
// opencv.ROI( faces[i].x, faces[i].y, faces[i].width, faces[i].height );
// opencv.brightness( 80 );
// opencv.contrast( 90 );
if(i==0)
{ image( fishImg,faces[i].x, faces[i].y, faces[i].width, faces[i].height); }
else if(i==1)
{ image( fishImg2,faces[i].x, faces[i].y, faces[i].width, faces[i].height); }
else if(i==2)
{ image( fishImg3,faces[i].x, faces[i].y, faces[i].width, faces[i].height); }
else if(i==3)
{ image( fishImg4,faces[i].x, faces[i].y, faces[i].width, faces[i].height); }
else if(i==4)
{ image( fishImg5,faces[i].x, faces[i].y, faces[i].width, faces[i].height); }
}
}
void extras(){
if(keyPressed){
if (key == 's' || key == 'S'){
flagForShark=1;
} else if(key=='n' || key =='N'){
flagForNotification=1;
} else if(key=='x' || key =='x'){
flagForNotification=0;
}
}
if(flagForShark==1){
// fill(255, 204, 255);
// stroke(128, 0, 128);
image( sharkImg,sharkMoves,sharkY);
//ellipse(candyX,candyY+candyMoves, 55, 55);
//image(loadImage("/Users/sang/Desktop/candy.png"),candyX,candyY+candyMoves);
if(sharkMoves>0){
sharkMoves-=sharkSpeed;
} else {
sharkMoves=480;
flagForShark=0;
}
}
if(flagForNotification==1){
image(sharkImg,NotificationX,NotificationY);
}
//Clock
// Draw the clock background
// fill(80);
noStroke();
// ellipse(cx, cy, clockDiameter, clockDiameter);
image(clockImg,5,5,clockDiameter,clockDiameter);
// Angles for sin() and cos() start at 3 o'clock;
// subtract HALF_PI to make them start at the top
float s = map(second(), 0, 60, 0, TWO_PI) - HALF_PI;
float m = map(minute() + norm(second(), 0, 60), 0, 60, 0, TWO_PI) - HALF_PI;
float h = map(hour() + norm(minute(), 0, 60), 0, 24, 0, TWO_PI * 2) - HALF_PI;
// Draw the hands of the clock
stroke(61,36,9);
strokeWeight(1);
line(cx, cy, cx + cos(s) * secondsRadius, cy + sin(s) * secondsRadius);
strokeWeight(2);
line(cx, cy, cx + cos(m) * minutesRadius, cy + sin(m) * minutesRadius);
strokeWeight(4);
line(cx, cy, cx + cos(h) * hoursRadius, cy + sin(h) * hoursRadius);
// Draw the minute ticks
// strokeWeight(2);
// beginShape(POINTS);
// for (int a = 0; a < 360; a+=6) {
// float x = cx + cos(radians(a)) * secondsRadius;
// float y = cy + sin(radians(a)) * secondsRadius;
// vertex(x, y);
// }
endShape();
//end of clock
}
Answer
The problem I see with your sketch is that you've got your analysis and rendering running at the same frame rate. In troubleshooting latency issues, I would propose a divide-and-conquer approach by building two sketches linked by OSC. OSC running on a single machine is quick enough that it won't introduce additional latency problems. Here's how I'd build the two sketches:
a processing sketch which takes image data in and does face tracking. This sketch would fire an OSC event each n seconds. Each event would be a list of locations to draw fish. Set this to 4 frames/sec to start.
a processing sketch which listens for OSC events from sketch 1 and renders the aquarium each n seconds. Set this to 15 frames/sec to start.
I've used this approach in a similar problem (tracking the brightest point within a frame, triggering sound events when point hits certain coordinates) and it paid to tune the analysis rate separately from the rendering rate.
Hope that helps - fire back if you have trouble with OSC.