particleCharacterization.py 7.18 KB
Newer Older
JosefBrandt's avatar
JosefBrandt committed
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
GEPARD - Gepard-Enabled PARticle Detection
Copyright (C) 2018  Lars Bittrich and Josef Brandt, Leibniz-Institut für 
Polymerforschung Dresden e. V. <bittrich-lars@ipfdd.de>    

This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.

This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with this program, see COPYING.  
If not, see <https://www.gnu.org/licenses/>.
"""

import numpy as np
import cv2
25
from copy import deepcopy
JosefBrandt's avatar
JosefBrandt committed
26

27 28 29
from .particleClassification.colorClassification import ColorClassifier
from .particleClassification.shapeClassification import ShapeClassifier
from segmentation import closeHolesOfSubImage
30
from errors import InvalidParticleError
31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47

class ParticleStats(object):
    longSize = None
    shortSize = None
    height = None
    area = None
    shape = None
    color = None

def particleIsValid(particle):
    if particle.longSize == 0 or particle.shortSize == 0:
        return False
    
    if cv2.contourArea(particle.contour) == 0:
        return False
    return True
    
JosefBrandt's avatar
JosefBrandt committed
48

49 50
def getParticleStatsWithPixelScale(cnt, fullimage, dataset):
    pixelscale = dataset.getPixelScale()
JosefBrandt's avatar
JosefBrandt committed
51
    
52
    newStats = ParticleStats()
53 54 55 56
    newStats.longSize, newStats.shortSize, newStats.area = getContourStats(cnt)
    newStats.longSize *= pixelscale
    newStats.shortSize *= pixelscale
    newStats.area *= (pixelscale**2)
57 58 59 60
    
    if 0 in [newStats.longSize, newStats.shortSize, newStats.area]:
        raise InvalidParticleError
    
61 62
    newStats.height = getParticleHeight(cnt, dataset)    
    newStats.shape = getParticleShape(cnt, newStats.height)
63
    
64 65
    partImg = getParticleImageFromFullimage(cnt, fullimage)
    newStats.color = getParticleColor(partImg)
66
    
67
    return newStats
68 69 70 71 72 73 74 75 76

def getParticleColor(imgRGB, colorClassifier=None):
    img = cv2.cvtColor(imgRGB, cv2.COLOR_RGB2HSV_FULL)
    meanHSV = cv2.mean(img)
    if colorClassifier is None:
        colorClassifier = ColorClassifier()
    color = colorClassifier.classifyColor(meanHSV)
    return color

77 78 79
def getParticleShape(contour, particleHeight, shapeClassifier=None):
    if shapeClassifier is None:
        shapeClassifier = ShapeClassifier()
JosefBrandt's avatar
JosefBrandt committed
80
    shape = shapeClassifier.classifyShape(contour, particleHeight)
81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
    return shape

def getParticleHeight(contour, dataset):
    zimg = getParticleImageFromFullimage(contour, dataset.getZvalImg())
    if zimg.shape[0] == 0 or zimg.shape[1] == 0:
        raise InvalidParticleError
        
    zimg = cv2.medianBlur(zimg, 5)
    avg_ZValue = np.mean(zimg[zimg > 0])
    if np.isnan(avg_ZValue):  #i.e., only zeros in zimg
        avg_ZValue = 0
    z0, z1 = dataset.zpositions.min(), dataset.zpositions.max()
    height = avg_ZValue/255.*(z1-z0) + z0
    return height

def getContourStats(cnt):
    ##characterize particle
        if cnt.shape[0] >= 5:       ##at least 5 points required for ellipse fitting...
            ellipse = cv2.fitEllipse(cnt)
            short, long = ellipse[1]
        else:
            rect = cv2.minAreaRect(cnt)
            long, short = rect[1]
            
        if short>long:
            long, short = short, long
        
        area = cv2.contourArea(cnt)
    
        return long, short, area

112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133
def mergeContours(contours):
    img, xmin, ymin, padding = contoursToImg(contours)
    return imgToCnt(img, xmin, ymin, padding)

def getParticleImageFromFullimage(contour, fullimage):
    contourCopy = deepcopy(contour)
    xmin, xmax, ymin, ymax = getContourExtrema(contourCopy)

    img = fullimage[ymin:ymax, xmin:xmax]
    mask = np.zeros(img.shape[:2])
    
    for i in range(len(contourCopy)):
        contourCopy[i][0][0] -= xmin
        contourCopy[i][0][1] -= ymin
        
    cv2.drawContours(mask, [contourCopy], -1, (255, 255, 255), -1)
    cv2.drawContours(mask, [contourCopy], -1, (255, 255, 255), 1)
    
    img[mask == 0] = 0
    img = np.array(img, dtype = np.uint8)
    return img
    
134
def contoursToImg(contours, padding=0):
135 136 137 138 139 140
    contourCopy = deepcopy(contours)
    xmin, xmax, ymin, ymax = getContourExtrema(contourCopy)
    
    padding = padding   #pixel in each direction
    rangex = int(np.round((xmax-xmin)+2*padding))
    rangey = int(np.round((ymax-ymin)+2*padding))
141 142
    if rangex == 0 or rangey == 0:
        raise InvalidParticleError
143 144 145 146 147 148 149 150 151 152 153

    img = np.zeros((rangey, rangex))
    for curCnt in contourCopy:
        for i in range(len(curCnt)):
            curCnt[i][0][0] -= xmin-padding
            curCnt[i][0][1] -= ymin-padding
        cv2.drawContours(img, [curCnt], -1, 255, -1)
        cv2.drawContours(img, [curCnt], -1, 255, 1)
    img = np.uint8(cv2.morphologyEx(img, cv2.MORPH_CLOSE, np.ones((3, 3))))
    return img, xmin, ymin, padding

154 155
def imgToCnt(img, xmin, ymin, padding=0):
    def getContour(img, flag):
156
        if cv2.__version__ > '3.5':
157
            contours, hierarchy = cv2.findContours(img, cv2.RETR_EXTERNAL, flag)
158
        else:
159
            temp, contours, hierarchy = cv2.findContours(img, cv2.RETR_EXTERNAL, flag)
160
        
161 162 163 164 165 166 167
        if len(contours) == 0:   #i.e., no contour found
            raise InvalidParticleError
        elif len(contours) == 1:   #i.e., exactly one contour found
            contour = contours[0]
        else:       #i.e., multiple contours found
            contour = getLargestContour(contours)
            
168 169
        return contour
    
170 171 172 173 174 175 176 177 178 179 180 181 182
    def getLargestContour(contours):
        areas = []
        for contour in contours:
            areas.append(cv2.contourArea(contour))
        maxIndex = areas.index(max(areas))
        print(f'{len(contours)} contours found, getting the largest one. Areas are: {areas}, taking contour at index {maxIndex}')
        return contours[maxIndex]

    img = closeHolesOfSubImage(img)
    contour = getContour(img, flag=cv2.CHAIN_APPROX_SIMPLE)

    if len(contour) < 5:
        contour = getContour(img, flag=cv2.CHAIN_APPROX_NONE)
183
    
184 185 186
    for i in range(len(contour)):
        contour [i][0][0] += xmin-padding
        contour [i][0][1] += ymin-padding
187
        
188
    return contour
189 190 191 192 193 194 195 196

def getContourExtrema(contours):
    try:
        cnt = np.vstack(tuple(contours))
        xmin, xmax = cnt[:,0,:][:, 0].min(), cnt[:,0,:][:, 0].max()
        ymin, ymax = cnt[:,0,:][:, 1].min(), cnt[:,0,:][:, 1].max()        
    except IndexError: #i.e., not a list of contours was passed, but an individual contour. Hence, the above indexing does not work
        xmin, xmax = cnt[:, 0].min(), cnt[:, 0].max()
197 198 199 200 201
        ymin, ymax = cnt[:, 1].min(), cnt[:, 1].max()
        
    xmin, xmax = int(round(xmin)), int(round(xmax))
    ymin, ymax = int(round(ymin)), int(round(ymax))
    
202 203
    return xmin, xmax, ymin, ymax

204 205 206 207 208 209 210 211 212
def getParticleCenterPoint(contour):
    img, xmin, ymin, padding = contoursToImg(contour)
    dist = cv2.distanceTransform(img, cv2.DIST_L2, 3)
    ind = np.argmax(dist)
    y = ind//dist.shape[1]-1
    x = ind%dist.shape[1]-1
    x += xmin
    y += ymin
    return x, y