[distance_factor_calculation] towards distance factor interpolation

This commit is contained in:
Xaver Roos 2022-04-29 11:51:48 +02:00
parent af9444ab26
commit 08a4abda62
3 changed files with 140 additions and 117 deletions

View File

@ -2,10 +2,22 @@ from turtle import left
import matplotlib.pyplot as plt
import numpy as np
from IPython import embed
from etrack import MarkerTask, ImageMarker
def crop_frame(frame, marker_positions):
# load the four marker positions
def mark_crop_positions(self):
task = MarkerTask("crop area", ["bottom left corner", "top left corner", "top right corner", "bottom right corner"], "Mark crop area")
im = ImageMarker([task])
marker_positions = im.mark_movie(file_name, frame_number)
print(marker_positions)
np.save('marker_positions', marker_positions)
return marker_positions
def assign_marker_positions(marker_positions):
bottom_left_x = marker_positions[0]['bottom left corner'][0]
bottom_left_y = marker_positions[0]['bottom left corner'][1]
bottom_right_x = marker_positions[0]['bottom right corner'][0]
@ -14,6 +26,21 @@ def crop_frame(frame, marker_positions):
top_left_y = marker_positions[0]['top left corner'][1]
top_right_x = marker_positions[0]['top right corner'][0]
top_right_y = marker_positions[0]['top right corner'][1]
return bottom_left_x, bottom_left_y, bottom_right_x, bottom_right_y, top_left_x, top_left_y, top_right_x, top_right_y
def assign_checkerboard_positions(checkerboard_marker_positions):
checkerboard_top_right = checkerboard_marker_positions[0]['top right corner']
checkerboard_top_left = checkerboard_marker_positions[0]['top left corner']
checkerboard_bottom_right = checkerboard_marker_positions[0]['bottom right corner']
checkerboard_bottom_left = checkerboard_marker_positions[0]['bottom left corner']
return checkerboard_top_right, checkerboard_top_left, checkerboard_bottom_right, checkerboard_bottom_left
def crop_frame(frame, marker_positions):
# load the four marker positions
bottom_left_x, bottom_left_y, bottom_right_x, bottom_right_y, top_left_x, top_left_y, top_right_x, top_right_y = assign_marker_positions(marker_positions)
# define boundaries of frame, taken by average of points on same line but slightly different pixel values
left_bound = int(np.mean([bottom_left_x, top_left_x]))
@ -22,12 +49,12 @@ def crop_frame(frame, marker_positions):
bottom_bound = int(np.mean([bottom_left_y, bottom_right_y]))
# crop the frame by boundary values
crop_frame = frame[top_bound:bottom_bound, left_bound:right_bound]
crop_frame = np.mean(crop_frame, axis=2) # mean over 3rd dimension (RGB/color values)
cropped_frame = frame[top_bound:bottom_bound, left_bound:right_bound]
cropped_frame = np.mean(cropped_frame, axis=2) # mean over 3rd dimension (RGB/color values)
# mean over short or long side of the frame corresponding to x or y axis of picture
frame_width = np.mean(crop_frame,axis=0)
frame_height = np.mean(crop_frame,axis=1)
frame_width = np.mean(cropped_frame,axis=0)
frame_height = np.mean(cropped_frame,axis=1)
# differences of color values lying next to each other --> derivation
diff_width = np.diff(frame_width)
@ -37,7 +64,7 @@ def crop_frame(frame, marker_positions):
x_width = np.arange(0, len(diff_width), 1)
x_height = np.arange(0, len(diff_height), 1)
return crop_frame, frame_width, frame_height, diff_width, diff_height, x_width, x_height
return cropped_frame, frame_width, frame_height, diff_width, diff_height, x_width, x_height
def rotation_angle():
pass

View File

@ -7,9 +7,10 @@ import numpy as np
import cv2
import os
import sys
import glob
from IPython import embed
from etrack import MarkerTask, ImageMarker
from calibration_functions import crop_frame, threshold_crossings, checkerboard_position, filter_data
from calibration_functions import *
class DistanceCalibration():
@ -17,24 +18,7 @@ class DistanceCalibration():
def __init__(self, file_name, frame_number, x_0=154, y_0=1318, cam_dist=1.36, tank_width=1.35, tank_height=0.805, width_pixel=1900, height_pixel=200,
checkerboard_width=0.24, checkerboard_height=0.18, checkerboard_width_pixel=500, checkerboard_height_pixel=350) -> None:
super().__init__()
# aktualisieren
"""Calibration of the dimensions of the tank. Conversion of pixel into meter. Width refers to the "x-axis", height to the "y-axis" of the tank.
Args:
file_name (_type_): _description_
x_0 (int, optional): X-value of the "origin" of the tank. Defaults to 0.
y_0 (int, optional): Y-value of the "origin" of the tank. Defaults to 0.
cam_dist (int, optional): Distance of camera lense to tank floor. Defaults to 1.36.
width (int, optional): Width in meter from one lightened corner of the tank to the other. Defaults to 1.35.
height (int, optional): Height in meter from one lightened corner of the tank to the other. Defaults to 1.35.
width_pixel (int, optional): Width in pixel from one lightened corner of the tank to the other. Defaults to 1975.
height_pixel (int, optional): Height in pixel from one lightened corner of the tank to the other. Defaults to 1375.
rectangle_width (float, optional): Width of one black or corresponding white rectangle of the checkerboard. Defaults to 0.024.
rectangle_height (float, optional): Height of one black or corresponding white rectangle of the checkerboard. Defaults to 0.0225.
rectangle_count_width (int, optional): Number of black rectangles over the width of the whole checkerboard. Defaults to 9.
rectangle_count_height (int, optional): Number of black rectangles over the height of the whole checkerboard. Defaults to 7.
"""
self._file_name = file_name
self._x_0 = x_0
self._y_0 = y_0
@ -50,24 +34,18 @@ class DistanceCalibration():
self._x_factor = tank_width / width_pixel # m/pix
self._y_factor = tank_height / height_pixel # m/pix
self.distance_calculation
self.distance_factor_calculation
self.mark_crop_positions
# if needed include setter: @y_0.setter def y_0(self, value): self._y_0 = value
@property
def x_0(self):
return self._x_0
@x_0.setter
def x_0(self, value):
self._x_0 = value
@property
def y_0(self):
return self._y_0
@y_0.setter
def y_0(self, value):
self._y_0 = value
@property
def cam_dist(self):
return self._cam_dist
@ -76,50 +54,26 @@ class DistanceCalibration():
def width(self):
return self._width
@width.setter
def width(self, value):
self._width = value
@property
def height(self):
return self._height
@height.setter
def height(self, value):
self._height = value
@property
def width_pix(self):
return self._width_pix
@width_pix.setter
def width_pix(self, value):
self._width_pix = value
@property
def height_pix(self):
return self._height_pix
@height_pix.setter
def height_pix(self, value):
self._height_pix_ = value
@property
def cb_width(self):
return self._cb_width
@cb_width.setter
def cb_width(self, value):
self._cb_width = value
@property
def cb_height(self):
return self._cb_height
@cb_height.setter
def cb_height(self, value):
self._cb_height = value
@property
def x_factor(self):
return self._x_factor
@ -127,22 +81,22 @@ class DistanceCalibration():
@property
def y_factor(self):
return self._y_factor
def mark_crop_positions(self):
task = MarkerTask("crop area", ["bottom left corner", "top left corner", "top right corner", "bottom right corner"], "Mark crop area")
im = ImageMarker([task])
marker_positions = im.mark_movie(file_name, frame_number)
print(marker_positions)
task = MarkerTask("crop area", ["bottom left corner", "top left corner", "top right corner", "bottom right corner"], "Mark crop area")
im = ImageMarker([task])
marker_positions = im.mark_movie(file_name, frame_number)
print(marker_positions)
np.save('marker_positions', marker_positions)
np.save('marker_positions', marker_positions)
return marker_positions
return marker_positions
def detect_checkerboard(self, filename, frame_number, marker_positions):
# load frame
if not os.path.exists(filename):
raise IOError("file %s does not exist!" % filename)
video = cv2.VideoCapture()
@ -150,11 +104,6 @@ class DistanceCalibration():
frame_counter = 0
success = True
frame = None
x_0 = self._x_0
y_0 = self._y_0
width_pix = self._width_pix
height_pix = self._height_pix
while success and frame_counter <= frame_number: # iterating until frame_counter == frame_number --> success (True)
print("Reading frame: %i" % frame_counter, end="\r")
@ -162,15 +111,23 @@ class DistanceCalibration():
frame_counter += 1
marker_positions = np.load('marker_positions.npy', allow_pickle=True) # load saved numpy marker positions file
bottom_left_marker = marker_positions[0]['bottom left corner']
bottom_right_marker= marker_positions[0]['bottom right corner']
top_left_marker = marker_positions[0]['top left corner']
top_right_marker = marker_positions[0]['top right corner']
# care: y-axis is inverted, top values are low, bottom values are high
croped_frame, frame_width, frame_height, diff_width, diff_height, _, _ = crop_frame(frame, marker_positions) # crop frame to given marker positions
cropped_frame, frame_width, frame_height, diff_width, diff_height, _, _ = crop_frame(frame, marker_positions) # crop frame to given marker positions
bottom_left_x = 0
bottom_left_y = np.shape(cropped_frame)[0]
bottom_right_x = np.shape(cropped_frame)[1]
bottom_right_y = np.shape(cropped_frame)[0]
top_left_x = 0
top_left_y = 0
top_right_x = np.shape(cropped_frame)[1]
top_right_y = 0
cropped_marker_positions = [{'bottom left corner': (bottom_left_x, bottom_left_y), 'top left corner': (top_left_x, top_left_y),
'top right corner': (top_right_x, top_right_y), 'bottom right corner': (bottom_right_x, bottom_right_y)}]
thresh_fact = 7 # factor by which the min/max is divided to calculate the upper and lower thresholds
# filtering/smoothing of data using kernel with n datapoints
@ -188,49 +145,60 @@ class DistanceCalibration():
# position of checkerboard in width
print('width..')
width_position, left_width_position, right_width_position = checkerboard_position(lci_width, uci_width)
# position of checkerboard in height
print('height..')
height_position, left_height_position, right_height_position = checkerboard_position(lci_height, uci_height) # left height refers to top, right height to bottom
if width_position == 'left' and height_position == 'left':
checkerboard_position_tank = 'top left'
elif width_position == 'left' and height_position == 'right':
checkerboard_position_tank = 'bottom left'
elif width_position == 'right' and height_position == 'right':
checkerboard_position_tank = 'bottom right'
elif width_position == 'right' and height_position == 'left':
checkerboard_position_tank = 'top right'
else:
checkerboard_position_tank = 'middle'
print(checkerboard_position_tank)
# final corner positions of checkerboard
top_left = np.array([left_width_position, left_height_position])
top_right = np.array([right_width_position, left_height_position])
bottom_left = np.array([left_width_position, right_height_position])
bottom_right = np.array([right_width_position, right_height_position])
checkerboard_marker_positions = [{'bottom left corner': (left_width_position, right_height_position), 'top left corner': (left_width_position, left_height_position),
'top right corner': (right_width_position, left_height_position), 'bottom right corner': (right_width_position, right_height_position)}]
print(top_left, top_right, bottom_left, bottom_right)
print(checkerboard_marker_positions)
# fig, ax = plt.subplots()
# ax.imshow(croped_frame)
# for p in top_left, top_right, bottom_left, bottom_right:
# ax.scatter(p[0], p[1])
# plt.show()
# embed()
# quit()
return top_left, top_right, bottom_left, bottom_right
checkerboard_top_right, checkerboard_top_left, checkerboard_bottom_right, checkerboard_bottom_left = assign_checkerboard_positions(checkerboard_marker_positions)
fig, ax = plt.subplots()
ax.imshow(cropped_frame)
for p in checkerboard_top_left, checkerboard_top_right, checkerboard_bottom_left, checkerboard_bottom_right:
ax.scatter(p[0], p[1])
ax.scatter(bottom_left_x, bottom_left_y)
ax.scatter(bottom_right_x, bottom_right_y)
ax.scatter(top_left_x, top_left_y)
ax.scatter(top_right_x, top_right_y)
plt.show()
return checkerboard_marker_positions, cropped_marker_positions, checkerboard_position_tank
def distance_calculation(self, top_left, top_right, bottom_left, bottom_right, marker_positions):
def distance_factor_calculation(self, checkerboard_marker_positions, marker_positions):
checkerboard_top_right, checkerboard_top_left, checkerboard_bottom_right, checkerboard_bottom_left = assign_checkerboard_positions(checkerboard_marker_positions)
checkerboard_width = 0.24
checkerboard_height = 0.18
checkerboard_width_pixel = top_right[0] - top_left[0]
checkerboard_height_pixel = bottom_right[1] - top_right[1]
checkerboard_width_pixel = checkerboard_top_right[0] - checkerboard_top_left[0]
checkerboard_height_pixel = checkerboard_bottom_right[1] - checkerboard_top_right[1]
x_factor = checkerboard_width / checkerboard_width_pixel
y_factor = checkerboard_height / checkerboard_height_pixel
bottom_left_x = marker_positions[0]['bottom left corner'][0]
bottom_left_y = marker_positions[0]['bottom left corner'][1]
bottom_right_x = marker_positions[0]['bottom right corner'][0]
bottom_right_y = marker_positions[0]['bottom right corner'][1]
top_left_x = marker_positions[0]['top left corner'][0]
top_left_y = marker_positions[0]['top left corner'][1]
top_right_x = marker_positions[0]['top right corner'][0]
top_right_y = marker_positions[0]['top right corner'][1]
bottom_left_x, bottom_left_y, bottom_right_x, bottom_right_y, top_left_x, top_left_y, top_right_x, top_right_y = assign_marker_positions(marker_positions)
tank_width_pixel = np.mean([bottom_right_x - bottom_left_x, top_right_x - top_left_x])
tank_height_pixel = np.mean([bottom_left_y - top_left_y, bottom_right_y - top_right_y])
@ -239,23 +207,51 @@ class DistanceCalibration():
tank_height = tank_height_pixel * y_factor
print(tank_width, tank_height)
return x_factor, y_factor
embed()
quit()
def distance_factor_interpolation(x_factors, y_factors):
pass
if __name__ == "__main__":
file_name = "/home/efish/etrack/videos/2022.03.28_3.mp4"
frame_number = 10
dc = DistanceCalibration(file_name=file_name, frame_number=frame_number)
# marker_positions = dc.mark_crop_positions()
top_left, top_right, bottom_left, bottom_right = dc.detect_checkerboard(file_name, frame_number=frame_number, marker_positions=np.load('marker_positions.npy', allow_pickle=True))
all_x_factor = []
all_y_factor = []
all_checkerboard_position_tank = []
for file_name in glob.glob("/home/efish/etrack/videos/*"):
# file_name = "/home/efish/etrack/videos/2022.03.28_4.mp4"
frame_number = 10
dc = DistanceCalibration(file_name=file_name, frame_number=frame_number)
dc.mark_crop_positions()
checkerboard_marker_positions, cropped_marker_positions, checkerboard_position_tank = dc.detect_checkerboard(file_name, frame_number=frame_number, marker_positions=np.load('marker_positions.npy', allow_pickle=True))
x_factor, y_factor = dc.distance_factor_calculation(checkerboard_marker_positions, marker_positions=cropped_marker_positions)
all_x_factor.append(x_factor)
all_y_factor.append(y_factor)
all_checkerboard_position_tank.append(checkerboard_position_tank)
x_factors = np.load('x_factors.npy')
y_factors = np.load('y_factors.npy')
all_checkerboard_position_tank = np.load('all_checkerboard_position_tank.npy')
dc.distance_calculation(top_left, top_right, bottom_left, bottom_right, marker_positions = np.load('marker_positions.npy', allow_pickle=True))
embed()
quit()
# next up: distance calculation with angle
# is this needed or are current videos enough?:
# laying checkerboard at position directly above and below / left and right to centered checkerboard near edge of tank
# calculating x and y factor for centered checkerboard, then for the ones at the edge
# --> afterwards interpolate between them to have continuous factors for whole tank
# maybe smaller object in tank to have more accurate factor
# make function to refine checkerboard detection at edges of tank by saying if no lower color values appears near edge --> checkerboard position then == corner of tank?
#
# mark_crop_positions why failing plot at end?
# with rectangles of checkerboard?

View File

@ -65,7 +65,7 @@ class ImageMarker:
self._fig.gca().set_title("All set and done!\n Window will close in 2s")
self._fig.canvas.draw()
plt.pause(2.0)
plt.close()
plt.close() #self._fig.gca().imshow(frame))
return [t.marker_positions for t in self._tasks]
def _key_press_event(self, event):