detect_music_chord.py

'''
Applies a fixed threshold to a blurred image of each video frame to create binary objects.
Draw bounding box around binary objects in each frame of video and saves location, area and aspect ratio for tracking.
To play music, download music_v1 and open GarageBand on MAC. Chose your instrumentation.  
Reset Kernal after each run. 

Use keyboard to adjust threshold. Click on image to enable keyboard.
Press 't' to select threshold. Press '+' and '-' to change value by increments of 1.
Hold shift while pressing '+' or '-' to change value by increments of 10. 
Press 'q' to quit

v2 09.02.2021 Uses keyboard to change variables
v1 08.31.2020

Ariana Nagainis CCC, San Francisco State University November 2021
Tom Zimmerman CCC, IBM Research March 2020
This work is funded by the National Science Foundation (NSF) grant No. DBI-1548297, Center for Cellular Construction.
Disclaimer:  Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.
'''
import numpy as np
import cv2
import keyboard as k # reads keyboard and updates program variable with key presses
import music_v1 as M #reads the music file
    

########## USER SETTINGS ##############################
vid='M6.mp4'    

detectFileName='M6.csv'      # output file containing object location, area, aspect ratio for each video frame
X_REZ=640; Y_REZ=480;               # viewing resolution
MIN_AREA=10                         # min area of object detected
MAX_AREA=1000                     # max area of object detected
THICK=1                             # bounding box line thickness
BLUR=7                              # object bluring to help detection
VGA=(640,480)
PROCESS_REZ=(320,240)
    
############# DEFINE VARIABLES ##################
detectHeader= 'FRAME,ID,X0,Y0,X1,Y1,XC,YC,AREA,AR,ANGLE'
FRAME=0; ID=1;  X0=2;   Y0=3;   X1=4;   Y1=5;   XC=6;   YC=7; AREA=8; AR=9; ANGLE=10; MAX_COL=11 # pointers to detection features
detectArray=np.empty((0,MAX_COL), dtype='int')  # detection features populated with object for each frame


################### MAIN ###################
print('''
How to use keyboard to adjust threshold.....
Click on image to enable keyboard.
Press 't' to select threshold.
Press '+' and '-' to change value by increments of 1.
Hold shift while pressing '+' or '-' to change value by increments of 10. 
Press 'q' to quit
''')

def getAR(obj):
    ((xc,yc),(w,h),(angle)) = cv2.minAreaRect(obj)  # get parameters from min area rectangle
    ar=0.0      # initialize aspect ratio as a floating point so calculations are done in floating point
    # calculate aspect ratio (always 1 or greater)
    if w>=h and h>0:
        ar=w/h
    elif w>0:
        ar=h/w
    return(xc,yc,ar,angle)         

######### start capturing frames of video #############
cap = cv2.VideoCapture(vid) 
#cap = cv2.VideoCapture(vid)         # start video file reader
frameCount=0                        # keeps track of frame number
while(cap.isOpened() and k.run):    # process each frame until end of video or 'q' key is pressed

    # check for key activity that changes program variables
    (thresh,b,x,y)=k.processKey()   # b,x,y variables currently not used, just put in there to show usage
    
    # get image
    ret, colorIM = cap.read()
    if not ret:                     # check to make sure there was a frame to read
        print('Can not find video or we are all done')
        break
    frameCount+=1
    
    # blur and threshold image
    colorIM=cv2.resize(colorIM,PROCESS_REZ)
    grayIM = cv2.cvtColor(colorIM, cv2.COLOR_BGR2GRAY)  # convert color to grayscale image       
    blurIM=cv2.medianBlur(grayIM,BLUR)                  # blur image to fill in holes to make solid object
    ret,binaryIM = cv2.threshold(blurIM,thresh,255,cv2.THRESH_BINARY_INV) # threshold image to make pixels 0 or 255
    
    # get contours  
    #dummy, contourList, hierarchy = cv2.findContours(binaryIM, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE) # all countour points, uses more memory
    contourList, hierarchy = cv2.findContours(binaryIM, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
    # draw bounding boxes around objects
    objCount=0 
    note_list = []                                     # used as object ID in detectArray
    for objContour in contourList:                  # process all objects in the contourList
        area = int(cv2.contourArea(objContour))     # find obj area        
        if area>MIN_AREA:                           # only detect large objects       
            PO = cv2.boundingRect(objContour)
            x0=PO[0]; y0=PO[1]; x1=x0+PO[2]; y1=y0+PO[3]
            cv2.rectangle(colorIM, (x0,y0), (x1,y1), (0,255,0), THICK) # place GREEN rectangle around each object, BGR
            log_area = int(np.log10(area)*100)
            print(area, log_area)
            note_list.append(log_area)
            objCount+=1 
    M.make_chord(note_list)# indicate processed an object
    #print('frame:',frameCount,'objects:',len(contourList),'big objects:',objCount)

    # shows results
    cv2.imshow('colorIM', cv2.resize(colorIM,VGA))      # display image

        
print ('Done')