We made it! Thanks for supporting Public Lab's 🎈 Mini Balloon & Kite Kit (part of Kickstarter Gold) -- watch the video!

Public Lab Wiki documentation


  • 4

Python Webcam Codes



This page is a draft to host some simple codes in order to speed up coding development in Python for the Infragram Webcam.

The codes presented here are are based on OpenCV 2 library for video processing and Numpy for array math (Numpy arrays are supported by OpenCV 2). Note: These codes were designed for the Infragram Webcam using Rosco #2007 filter, which means that the blue channel contains blue light information, while the red channel contains red/infrared.

Basic Functions

NDVI function

The color code for this function is different from the one used at infragram.org so to have a smooth transition from negative NDVI values to positive NDVI values (on the website it toggles from cyan to green). In this version, the color scale is smoother, which hides noise near zero NDVI.

import numpy as np #arrays and math
import cv2 #opencv library

#-------------------------------------------
#----------------NDVI Function--------------
#-------------------------------------------

#NDVI Calculation
#Input: an RGB image frame from infrablue source (blue is blue, red is pretty much infrared)
#Output: an RGB frame with equivalent NDVI of the input frame
def NDVICalc(original):
    "This function performs the NDVI calculation and returns an RGB frame)"
    lowerLimit = 5 #this is to avoid divide by zero and other weird stuff when color is near black

    #First, make containers
    oldHeight,oldWidth = original[:,:,0].shape; 
    ndviImage = np.zeros((oldHeight,oldWidth,3),np.uint8) #make a blank RGB image
    ndvi = np.zeros((oldHeight,oldWidth),np.int) #make a blank b/w image for storing NDVI value
    red = np.zeros((oldHeight,oldWidth),np.int) #make a blank array for red
    blue = np.zeros((oldHeight,oldWidth),np.int) #make a blank array for blue

    #Now get the specific channels. Remember: (B , G , R)
    red = (original[:,:,2]).astype('float')
    blue = (original[:,:,0]).astype('float')

    #Perform NDVI calculation
    summ = red+blue
    summ[summ<lowerLimit] = lowerLimit #do some saturation to prevent low intensity noise

    ndvi = (((red-blue)/(summ)+1)*127).astype('uint8')  #the index

    redSat = (ndvi-128)*2  #red channel
    bluSat = ((255-ndvi)-128)*2 #blue channel
    redSat[ndvi<128] = 0; #if the NDVI is negative, no red info
    bluSat[ndvi>=128] = 0; #if the NDVI is positive, no blue info


    #And finally output the image. Remember: (B , G , R)
    #Red Channel
    ndviImage[:,:,2] = redSat

    #Blue Channel
    ndviImage[:,:,0] = bluSat

    #Green Channel
    ndviImage[:,:,1] = 255-(bluSat+redSat)

    return ndviImage;

DVI Function

This is the non-normalized version. Most of the times it has no significant difference from NDVI, but for people interested in it here it is.

import numpy as np #arrays and math
import cv2 #opencv library

#-------------------------------------------
#----------------DVI Function---------------
#-------------------------------------------

#DVI Calculation
#Input: an RGB image frame from infrablue source (blue is blue, red is pretty much infrared)
#Output: an RGB frame with equivalent DVI of the input frame
def DVICalc(original):
    "This function performs the DVI calculation and returns an RGB frame)"

    #First, make containers
    oldHeight,oldWidth = original[:,:,0].shape; 
    dviImage = np.zeros((oldHeight,oldWidth,3),np.uint8) #make a blank RGB image
    dvi = np.zeros((oldHeight,oldWidth),np.int) #make a blank b/w image for storing DVI value
    red = np.zeros((oldHeight,oldWidth),np.int) #make a blank array for red
    blue = np.zeros((oldHeight,oldWidth),np.int) #make a blank array for blue

    #Now get the specific channels. Remember: (B , G , R)
    red = (original[:,:,2]).astype('float')
    blue = (original[:,:,0]).astype('float')

    #Perform DVI calculation
    dvi = (((red-blue)+255)/2).astype('uint8')  #the index

    redSat = (dvi-128)*2  #red channel
    bluSat = ((255-dvi)-128)*2 #blue channel
    redSat[dvi<128] = 0; #if the NDVI is negative, no red info
    bluSat[dvi>=128] = 0; #if the NDVI is positive, no blue info


    #And finally output the image. Remember: (B , G , R)
    #Red Channel
    dviImage[:,:,2] = redSat

    #Blue Channel
    dviImage[:,:,0] = bluSat

    #Green Channel
    dviImage[:,:,1] = 255-(bluSat+redSat)

    return dviImage;

Example Code

Using the above functions (not pasted in the following code for brevity), a simple code to output the NDVI, DVI and Raw stream from webcam is presented. A snapshot function is also available.

#-------------------------------------------
#----------------Main Function--------------
#-------------------------------------------

cv2.namedWindow("preview NDVI")
vc = cv2.VideoCapture(0)

if vc.isOpened(): # try to get the first frame
    rval, frame = vc.read()
    height = vc.get(3) #get height
    width = vc.get(4) #get width
    #Text Related
    x = int(width/2)
    y = int(2*height/3) 
    text_color = (255,255,255) #color as (B,G,R)
    font = cv2.FONT_HERSHEY_PLAIN
    thickness = 2
    font_size = 2.0
else:
    rval = False

while rval:
    ndviImage = NDVICalc(frame)
    dviImage = DVICalc(frame)

    cv2.putText(frame, "Raw Image", (x,y), font, font_size, text_color, thickness, lineType=cv2.CV_AA)
    cv2.putText(ndviImage, "NDVI Image", (x,y), font, font_size, text_color, thickness, lineType=cv2.CV_AA)
    cv2.putText(dviImage, "DVI Image", (x,y), font, font_size, text_color, thickness, lineType=cv2.CV_AA)

    newFrame = np.concatenate((ndviImage,dviImage,frame),axis=1)
    cv2.imshow("preview NDVI", newFrame)

    rval, frame = vc.read()

    key = cv2.waitKey(1)&0xFF #get a key press
    if key == ord('q'): #q for quitting
        break
    elif key == ord('p'): #p for printscreen
        curtime = datetime.datetime.now()
        formattedTime = curtime.strftime("%Y%m%d-%H-%M-%S.jpg")
        print 'filename:%s'%formattedTime
        cv2.imwrite(formattedTime,newFrame)
        print "Screenshot taken!"

# When everything done, release the capture
vc.release()
cv2.destroyAllWindows()

software infragram python

parent:multispectral-imaging