Files
Ivolution/facemovie/Facemovie.py

395 lines
14 KiB
Python

"""
.. module:: Facemovie
:platform: Unix, Windows
:synopsis: Main class of the application. Contains the core image processing
functions.
Plays the role of a controller for the application, as it supports the
communication layer with the end user interface.
.. moduleauthor:: Julien Lengrand-Lambert <jlengrand@gmail.com>
"""
import os
import sys
import cv
from facemovie.lib import exif
from facemovie import Guy
class FaceMovie(object):
'''
Main class of the whole application.
Contains the core image processing functions.
Takes a bunch of parameters and a list of images and tries to create a
video out of it.
Contains general methods, aimed at being used trough an interface.
'''
def __init__(self, in_folder, out_folder, face_params):
"""
Initializes all parameters of the application. Input and output folders
are defined, together with the classifier profile.
Args:
in_folder (string) : the location where input files will be
searched
out_folder (string) : the location where the outputs will be
saved
face_param (string) : the location of the profile file used
to train the classifier
"""
self.source= in_folder # Source folder for pictures
self.out = out_folder # Folder to save outputs
self.guys = [] # List of pictures in source folder
# Retrieving parameters for Face Detection
self.face_params = face_params
# Position of the center in output images
self.x_center = 0
self.y_center = 0
# minimum size needed on right of center
self.x_af = 0
self.y_af = 0
# Needed minimum size of output image
self.dim_x = 0
self.dim_y = 0
# thumbmails
self.crop = False
self.cropdims = [0, 0] # user defined desired dimensions for cropping
self.width = [0, 0]
self.height = [0, 0]
self.face_mean = [0, 0]
self.sort_method = "n" # sorting by name or using metadata (n or e)
def set_crop_dims(self, crop_x, crop_y):
"""
Sets the cropping dimension in case they have been provided by the end user
Args:
crop_x (int) : dimension of the desired cropping in x (in number of face size)
crop_y (int) : dimension of the desired cropping in y (in number of face size)
"""
self.cropdims = [crop_x, crop_y]
def list_guys(self):
"""
Aims at populating the guys list, using the source folder as an input.
Guys list shall be sorted chronologically.
In case no valid date is found, it is set to ''.
"""
try:
os.path.exists(self.source)
os.path.isdir(self.source) # checking if folder exists
except : # find precise exception
print "ERROR : Source folder not found ! Exiting. . ."
sys.exit(0)
# just listing directory. Lets be more secure later
files = os.listdir(self.source)
# loading images, create Guys and store it into guys
for token in files :
image = cv.LoadImage(os.path.join(self.source, token))
guy_name = os.path.splitext(token)[0]
try:
path = os.path.join(self.source, token)
guy_date = exif.parse(path)['DateTime']
except Exception:
guy_date = ''
a_guy = Guy.Guy(image, guy_name, guy_date)
# populating guys
self.guys.append(a_guy)
# Sorting either by exif date or name
if self.sort_method == "e":
print "Sorting files using EXIF metadata"
self.guys.sort(key=lambda g: g.date)
else: # default is sort by name
print "Sorting files using file name"
self.guys.sort(key=lambda g: g.name)
def search_faces(self):
"""
Searches for all faces in the guys we have
Results to be stored directly in guys
Takes each image one after the other, and create a guy out of it.
The Face of each guy is searched.
In case no face is found, a warning is returned and Guy is set to None
"""
for a_guy in self.guys:
a_guy.search_face(self.face_params)
if a_guy.has_face(): # face(s) have been found
print "%d faces found for %s" % (a_guy.num_faces(), a_guy.name)
else:
print "Warning! No face found for %s" %(a_guy.name)
def normalize_faces(self, reference=0):
"""
Creates new images, normalized by face size
A reference is given in input. The idea is to get all images to have the
same size in Guy.
KArgs:
reference (int) : the reference size of the face that we want to have
"""
# FIXME: May be enhanced by choosing a more educated reference
if reference == 0:
reference = self.guys[0].faces[0][0][3] # catch face size (width)
for a_guy in self.guys:
if a_guy.has_face():
a_guy.normalize_face(reference)
def calc_mean_face(self):
"""
Returns the mean size of all faces in input
Used to correctly crop images
**Designed for internal use only**
"""
tot_x = 0
tot_y = 0
nb_face = 0
for a_guy in self.guys:
if a_guy.has_face():
((x, y, w, h), n) = a_guy.faces[0]
tot_x += w
tot_y += h
nb_face += 1
self.face_mean = [float(tot_x) / nb_face, float(tot_y) / nb_face]
def find_crop_dims(self):
"""
Calculates smallest output image that can be used to avoid adding black borders on image
It will later be used to create the final image.
The idea is the same as for :func:find_out_dims , but while avoiding black brders.
"""
ht = 1000000 # space left above eyes
hb = 1000000 # space left beneath eyes
wl = 1000000 # space left left of eyes
wr = 1000000 # space left right of eyes
if self.cropdims != [0, 0]:
w = int( (self.cropdims[0] * self.face_mean[0]) / 2)
self.width = [w, w]
h = int((self.cropdims[1] * self.face_mean[1]) / 2)
self.height = [h, h]
else:
for a_guy in self.guys:
if a_guy.has_face():
xc = a_guy.x_center
yc = a_guy.y_center
inx = a_guy.in_x
iny = a_guy.in_y
# finding width
if xc < wl:
wl = xc
if (inx - xc) < wr:
wr = inx - xc
# finding height
if yc < ht:
ht = yc
if (iny - yc) < hb:
hb = iny - yc
self.width = [wl, wr]
self.height = [ht, hb]
if (sum(self.width) >= self.dim_x) or (sum(self.height) >= self.dim_y):
print "Cropping inactive : Maximum dimensions reached"
self.crop = False
else:
self.crop = True
def find_out_dims(self):
"""
Calculates best output image size and position depending on
faces found in guys.
The system is simple. The output image should be as big as possible,
and faces are always placed in the same position. Depending on that,
the image input image is placed in the output at the correct position.
Black borders are set everywhere else.
"""
# FIXME: badly done !
for a_guy in self.guys:
if a_guy.has_face():
xc = a_guy.x_center
yc = a_guy.y_center
inx = a_guy.in_x
iny = a_guy.in_y
# update center
if xc > self.x_center:
self.x_center = xc
if yc > self.y_center:
self.y_center = yc
# update right part
if (inx - xc) > self.x_af:
self.x_af = inx - xc
if (iny - yc) > self.y_af:
self.y_af = iny - yc
self.dim_x = self.x_af + self.x_center
self.dim_y = self.y_af + self.y_center
# finishes by calculating average face size
self.calc_mean_face()
def crop_im(self, image):
"""
If needed, crops the image to avoid having black borders.
Args:
image (IplImage) : the image to be cropped
"""
# TODO : implement
width = self.width#[0, 0]
height = self.height#[0, 0]
out_im = cv.CreateImage((width[0] + width[1], height[0] + height[1]),cv.IPL_DEPTH_8U, image.nChannels)
cv.Zero(out_im)
xtl = self.x_center - width[0]
ytl = self.y_center - height[0]
w = width[0] + width[1]
h = height[0] + height[1]
rect = (xtl, ytl, w, h)
cv.SetImageROI(image, rect)
cv.Copy(image, out_im)
cv.ResetImageROI(image)
return out_im
def show_faces(self, mytime=1000):
"""
Show all faces that have been found for the guys.
The time for which each image will be displayed can be chosen.
KArgs :
mytime (int) : time for which the image should be displayed (in ms)
"""
for a_guy in self.guys:
if a_guy.has_face():
out_im = a_guy.create_video_output(self.dim_x,
self.dim_y,
self.x_center,
self.y_center)
if self.crop:
out_im = self.crop_im(out_im)
self.out_display(out_im, a_guy.name, time=mytime)
def save_faces(self, out_folder, im_format="png"):
"""
Save all faces into out_folder, in the given image format
Args:
out_folder (string) : the location where to save the output image.
KArgs :
mytime (int) : time for which the image should be displayed (in ms)
"""
for a_guy in self.guys:
if a_guy.has_face():
out_im = a_guy.create_video_output(self.dim_x,
self.dim_y,
self.x_center,
self.y_center)
if self.crop:
out_im = self.crop_im(out_im)
self.save_result(out_im, a_guy.name, out_folder, im_format)
def save_movie(self, out_folder, fps=3):
"""
Creates a movie with all faces found in the inputs.
Guy is skipped if no face is found.
Args:
out_folder (string) : the location where to save the output image.
KArgs :
fps (int) : the number of frames per second to be displayed in final video
"""
filename = os.path.join(out_folder, "output.avi")
fourcc = cv.CV_FOURCC('C', 'V', 'I', 'D')
if self.crop:
width = self.width
height = self.height
frameSize = (width[0] + width[1], height[0] + height[1])
else:
frameSize = (self.dim_x, self.dim_y)
my_video = cv.CreateVideoWriter(filename,
fourcc,
fps,
frameSize,
1)
ii = 0
for a_guy in self.guys:
ii += 1
if a_guy.has_face():
print "frame %d" %(ii)
out_im = a_guy.create_video_output(self.dim_x,
self.dim_y,
self.x_center,
self.y_center)
if self.crop:
out_im = self.crop_im(out_im)
cv.WriteFrame(my_video, out_im)
def number_guys(self):
"""
Simply returns the number of guys in the current to-be movie
__Designed for interface use only__
"""
return len(self.guys)
def out_display(self, im, name, time=1000, im_x=640, im_y=480):
"""
Displays the output image, for time ms.
Setting time to 0 causes the image to remains open.
Window name slightly changed to match output
Args:
im (IplImage) : the image to be saved, formatted as an OpenCV Image
name (string) : the name of the image to be saved
KArgs :
time (int) : time for which the image should be displayed (in ms)
im_x (int) : output size of the displayed image (in pixels)
im_y (int) : output size of the displayed image (in pixels)
"""
win_name = name + " - out"
cv.NamedWindow(win_name, cv.CV_WINDOW_NORMAL)
cv.ResizeWindow(win_name, im_x, im_y)
cv.ShowImage(win_name, im)
cv.WaitKey(time)
cv.DestroyWindow(win_name)
def save_result(self, im, name, out_folder, ext):
"""
Saves output image to the given format (given in extension)
Args:
im (IplImage) : the image to be saved, formatted as an OpenCV Image
name (string) : the name of the image to be saved
out_folder (string) : the location where to save the image
ext (string) : Format in which the image should be saved
"""
file_name = name + "." + ext
out_name = os.path.join(out_folder, file_name)
print "Saving %s" %(out_name)
cv.SaveImage(out_name, im)