''' Written by Steve Flanagan, 2022 contact: saf204@gmail.com Code to extract images from the FireCams built by Dan Julio ref: https://github.com/danjulio/firecam Can run through all cameras at once if placed in a folder above where all the ".json" files are stored Creates a folder "ExtractedImages" with a sub-folder for each camera Each camera folder will have then have sub-folders "OriginalImages" which are the extracted visual and IR images, and "CSVfiles" which are the IR x,y coordinates needed for the IR_processing script If the user wants, they can turn on "additional_images" and "movie" in the inputs below. Additional folders will then be made that produce images and a movie that might help the user when they run the IR_processing script. ''' ################################################################# # START USER INPUT ################################################################# #Enter the file path to the '.json' files produced by the Firecams #wdir='C:/users/RxLabComp1/Desktop/Ft_Stewart_2024/FireOnly/serdp_img/' wdir='C:/users/RxLabComp1/Desktop/Ft_Stewart_2024/2025_burns/FireOnly/combined_for_wfsi/' #wdir='C:/users/RxLabComp1/Desktop/Ft_Stewart_2024/2025_burns/FireOnly/BU8/' #wdir='C:/users/RxLabComp1/Desktop/gallagher_data/card1_cameraFF39/SDHC/' #wdir = "/mnt/c/Users/sflanagan/Desktop/thermal_camera/trial/" #Ubuntu #Filter adjustments, Set nd_mult to 1 and nd_ladj is zero if a filter isn't used nd_mult = 7.1 #the "m" in y=mx + b nd_ladj = -136.4 #the "b" #Metadata should store ambient temperatures, and the code pulls that in for an adjustment #In the rare case where the meta data fails to record temp, this is used as the default #In Celcius amb_temp = 25 ########################OPTIONAL INPUTS########################### #Set to 0 if you do not want them, 1 if you do #Additional images are primarily used for comparing visual to IR additional_images = 0 #Set to 1 if you want to make additional images movie = 0 #Set to 1 if you want to make a move, #additional_images nust also be 1 #Temperatire scale for additional IR images - Kelvin min_temp = 300 max_temp = 1200 ################################################################# # END USER INPUT # ################################################################# #libraries import os, shutil, sys, base64, json, gc import numpy as np import matplotlib.pyplot as plt import PIL from PIL import Image import matplotlib matplotlib.use('agg') from datetime import datetime if movie==1: import cv2 os.chdir(wdir) # Get a list of all json files in the folder and subfolders # Check that files are found, if not, exit # Check that data hasn't already been processed # If it has, ask the user if they would like to reprocess, otherwise exit #Check if files already processed print("Making a directory for the extracted images") try: os.mkdir(wdir+"ExtractedImagesNewCalFinal/") except: print("Directory already exists") overwrite = input("To overwrite, enter 'y'") if overwrite == "y": print("Removing directory, Reprocessing images") shutil.rmtree(wdir+"ExtractedImagesNewCalFinal/") os.mkdir(wdir+"ExtractedImagesNewCalFinal/") else: print('Exiting script') sys.exit(1) #Get a list of cameras obj_in_wdir = os.scandir(wdir) cameras=[] for entry in obj_in_wdir: if entry.is_dir(): cameras.append(entry.name) cameras = [x for x in cameras if not x.startswith('Ext')] print("Cameras found to process:") print(cameras) for c in range(0,len(cameras)): #for c in range(2,3): filelist = [] print("Processing camera " +str(c+1) +" of " + str(len(cameras))) print("Checking for 'json' files in " +cameras[c] +" folder and sub-folders") #Check for json files, list if found, exit otherwise for r, d, f in os.walk(wdir+cameras[c]+"/"): for file in f: if '.json' in file: filelist.append(os.path.join(r,file).replace("\\","/")) if len(filelist) > 0: print(str(len(filelist))+" .json files found for " +cameras[c]) else: print("No 'json' files found, check that file structure is correct") print("Exiting script") sys.exit(1) # Output Directories print("Creating output dirs") ndir = wdir+"ExtractedImagesNewCalFinal/"+cameras[c]+"/" os.makedirs(ndir) os.chdir(ndir) os.makedirs("OriginalImages") os.makedirs("CSVfiles") if additional_images ==1: os.makedirs("ResizedImages") os.makedirs("OverlayedImages") os.makedirs("CombinedImages") if movie == 1: os.makedirs("IRandVisualOnly") #Error log file errorlog = open("errorlog.txt", 'w+') errorlog.write("Camera Type Failed Image \n") # Image failure checks and total count radfail = 0 jpegfail = 0 count =0 # Extract images for each json file print("Every failed rad or jpeg will print to the screen") print("And an update of every 1000 files processed") for f in filelist: # Check for if either image failed - if they did, don't do extra images imgFail = 0 #Open file myjsonfile = json.load( open(f) ) outputname = f.split('/')[-1].split('.json')[0] cameraName = myjsonfile['metadata']['Camera'] if count ==0: #camera sees itself. baseline copies first image #and then every image after this subtracts that value #to perform the correction. img_array=np.zeros((120,160,len(filelist)-1)) baseline=np.zeros((120,160)) try: temp_adj = myjsonfile['metadata']['Lens Temp'] except: temp_adj = amb_temp #x,y,t array parameters time_array = np.zeros((len(filelist)-1)) time_start = myjsonfile["metadata"]["Time"] date_format = '%H:%M:%S' ts_converted = datetime.strptime(time_start,date_format) if count%1000==0: print('Processing image ' + str(count) + " of " + str(len(filelist))+" for " +cameras[c]) # Decode the jpeg data and save the visual image try: jpegbinary = base64.b64decode( myjsonfile['jpeg'] ) o = open("OriginalImages/"+outputname+"_visual.jpg", 'wb') o.write(jpegbinary) o.close() except: print("jpeg failed " + f) errorlog.write(cameraName +" Visible " +f.split('/')[-1].split('.json')[0] + "\n") jpegfail += 1 imgFail = 1 # Decode the radiometric data try: radbinary = base64.b64decode( myjsonfile['radiometric'] ) raddata2D = np.arange(19200).reshape(120,160) raddata2D= np.frombuffer(radbinary,dtype=np.uint16).reshape(raddata2D.shape) # Convert it based on gain used to C final_array = np.zeros((120,160)) try: res = float(myjsonfile["metadata"]["Lepton Resolution"]) except: res = 0.01 #default is high gain if res == 0.1: final_array = (raddata2D - 2731.50)/10.00*float(nd_mult)+float(nd_ladj) else: final_array = (raddata2D - 27315.0)/100.0*float(nd_mult)+float(nd_ladj) if count==0: baseline = final_array else: final_array-=baseline final_array+=temp_adj+273.15 #Save arrays if count != 0: #set missfires to ambient temp final_array[final_array>3000]=temp_adj+273.15 img_array[:,:,count-1]=final_array #get current time and difference from start current_time = myjsonfile["metadata"]["Time"] ct_converted = datetime.strptime(current_time,date_format) diff = ct_converted-ts_converted diff_int = int(diff.total_seconds()) time_array[count-1]=diff_int # Save as CSV for data processing outputcsv = f.split('/')[-1].split('.json')[0]+'_IR.csv' np.savetxt("CSVfiles/"+outputcsv, final_array.astype(float), delimiter=',', header = myjsonfile["metadata"]["Time"]) # Save as image outputrad = f.split('/')[-1].split('.json')[0]+'_IR.png' #plt.imsave("OriginalImages/"+outputrad, final_array, cmap = 'RdBu_r', vmin = min_temp, vmax= max_temp) plt.imshow(final_array, cmap = 'RdBu_r') plt.colorbar() plt.title('Temperature (K)') plt.savefig("OriginalImages/"+outputrad) plt.cla() plt.clf() plt.close('all') gc.collect() except: print('rad failed ' + f ) errorlog.write(cameraName +" Radiometric " + f.split('/')[-1].split('.json')[0] +"\n") radfail +=1 imgFail = 1 count +=1 # If both images were recorded, perform some comparisons if imgFail == 0 and additional_images==1: # Resize the visual pixels (640 x 480) to matcth the radiometric (160x120) jpeg_resized = np.array(Image.open("OriginalImages/"+outputname+"_visual.jpg").resize((160,120))) plt.imsave("ResizedImages/"+outputname+".png",jpeg_resized) #plt.savefig(ndir+"/ResizedImages/"+outputjpeg) plt.cla() plt.clf() plt.close('all') gc.collect() # Overlay the two images plt.imshow(final_array, cmap='tab20c' , vmin = min_temp, vmax= max_temp) plt.colorbar() plt.imshow(jpeg_resized,alpha=.7) plt.savefig("OverlayedImages/"+outputname) plt.cla() plt.clf() plt.close('all') gc.collect # Combine visual, IR, overlayed, and a grid into one image fig, ax = plt.subplots(nrows=2, ncols=2, clear=True) ax[0,0].imshow(final_array,cmap = 'RdBu_r', vmin = min_temp, vmax= max_temp) ax[0,1].imshow(jpeg_resized) ax[1,1].imshow(jpeg_resized) ax[1,1].imshow(final_array, cmap='RdBu_r' ,alpha = .7, vmin = min_temp, vmax= max_temp) ax[1,0].set_xticks(np.arange(0,160,5)) ax[1,0].set_yticks(np.arange(0,120,5)) ax[1,0].grid(color='black') ax[1,0].imshow(final_array,cmap = 'RdBu_r', vmin = min_temp, vmax= max_temp) plt.savefig("CombinedImages/"+outputname) ax[1,1].cla() ax[0,0].cla() ax[1,0].cla() ax[0,1].cla() plt.clf() plt.close('all') plt.close(fig) gc.collect() #For movie if movie==1: fig, ax = plt.subplots(nrows=1, ncols=2, clear=True) ax[0].axis('off') ax[0].set_title('IR') ax[0].imshow(final_array,cmap = 'RdBu_r', vmin = min_temp, vmax= max_temp) ax[1].axis('off') ax[1].set_title('Visual') ax[1].imshow(jpeg_resized) plt.savefig("IRandVisualOnly/"+outputname) ax[0].cla() ax[1].cla() plt.clf() plt.close('all') plt.close(fig) gc.collect() #Make a movie of IR and visual combined if movie ==1: #os.chdir(ndir+cameras[c]) image_folder = 'IRandVisualOnly' video_name = 'IRandVisual.avi' images = [img for img in os.listdir(image_folder) if img.endswith(".png")] frame = cv2.imread(os.path.join(image_folder, images[0])) height, width, layers = frame.shape video = cv2.VideoWriter(video_name, 0, 1, (width,height)) for image in images: video.write(cv2.imread(os.path.join(image_folder, image))) cv2.destroyAllWindows() video.release() #Write out total failures and close script errorlog.write("\n\n\n Total camera failues: \n Visible "+str(jpegfail)+"\n Radiometric " +str(radfail)) errorlog.close() img_array[0,0,:] = time_array np.save(cameras[c]+"_array",img_array) #endtime = datetime.datetime.now() #print("end time " + endtime.strftime("%Y-%m-%d %H:%M:%S")) #print("start time " + starttime.strftime("%Y-%m-%d %H:%M:%S")) print('End of script') sys.exit(1)