#################################################
allFiles = next(os.walk("../TrainingSet/Offline Forgeries"))[1]
dataToTrain = {}
for files in allFiles:
folderFiles = next(os.walk("../TrainingSet/Offline Forgeries/" +
files))[2]
for signature in folderFiles:
img = rz.resizeSignature("../TrainingSet/Offline Forgeries/" +
files + "/" + signature)
# Image filtering
imgSmooted = GaussS.gaussianBlur(1, 0.55, img)
imgFiltered = bN.filter(imgSmooted)
# Calculate the ones in the matrix.
blackPoints = bN.calculateBlackPoints(imgFiltered)
# Calculate the center of mass
centerMass = bN.centroid(blackPoints)
# Calculate the eccentricity
eccentricity = bN.calculateEccentricity(blackPoints)
# Calculate the representative points of a signature
# densePoints = bN.calculateDensePoints(blackPoints)
# Calculate Kurtosis of blackpoints in signature
kurtosis = bN.calculateKurtosis(blackPoints)
def loadImage(self):
bashCommand = "rm test.PNG"
import subprocess
process = subprocess.Popen(bashCommand.split(), stdout=subprocess.PIPE)
output, error = process.communicate()
print output
# first of all, the base transformation of the data points is needed
base = plt.gca().transData
rot = transforms.Affine2D().rotate_deg(270)
filename = askopenfilename()
img = rz.resizeSignature(filename)
# Image filtering
imgSmooted = GaussS.gaussianBlur(1, 0.5, img)
imgFiltered = bN.filter(imgSmooted)
# Calculate the ones in the matrix.
blackPoints = bN.calculateBlackPoints(imgFiltered)
# Calculate the center of mass
centerMass = bN.centroid(blackPoints)
self.CenterMass.setText(str(centerMass[1]))
self.CenterMass_2.setText(str(centerMass[0]))
# Calculate the eccentricity
eccentricity = bN.calculateEccentricity(blackPoints)
self.Eccentricity.setText(str(round(eccentricity,4)))
# Calculate the representative points of a signature
densePoints = bN.calculateDensePoints(blackPoints)
listY = [y[0] for y in blackPoints]
listX = [x[1] for x in blackPoints]
plt.plot(listY, listX, "ro", transform=rot + base)
# Print Center of mass
plt.plot(centerMass[0], centerMass[1], "^", transform=rot + base)
# Print dense points
densePoints = densePoints.T
plt.plot(densePoints[0], densePoints[1], "g+", transform=rot + base)
kurtosis = bN.calculateKurtosis(blackPoints)
self.kurtosis.setText(str(round(kurtosis[1], 4 )))
self.kurtosis_2.setText(str(round(kurtosis[0], 4)))
# Print skew detection
testSkew = bN.skewDetection(imgFiltered)
self.Skew.setText(str(round(testSkew)))
# Calculate gradient in a grid 4x4 - Choosen just 2x2 important grid in the middle
matrixGradient = bN.calculateGradient(blackPoints, imgFiltered.shape)
self.Gradient.setRowCount(len(matrixGradient))
self.Gradient.setColumnCount(len(matrixGradient[0]))
for i, row in enumerate(matrixGradient):
for j, val in enumerate(row):
self.Gradient.setItem(i, j, QtGui.QTableWidgetItem(str(val)))
self.Gradient.resizeColumnsToContents()
self.Gradient.resizeRowsToContents()
# Calculate pressure in a grid 4x4
matrixPresure = np.around(bN.calculatePressure(imgSmooted), decimals=2)
self.Presure.setRowCount(len(matrixPresure))
self.Presure.setColumnCount(len(matrixPresure[0]))
for i, row in enumerate(matrixPresure):
for j, val in enumerate(row):
self.Presure.setItem(i, j, QtGui.QTableWidgetItem(str(val)))
self.Presure.resizeColumnsToContents()
self.Presure.resizeRowsToContents()
plt.axis('off')
# Plot Ellipse
subplot = plt.subplot()
b, a = drawEllipse(blackPoints)
ell = Ellipse((centerMass[1], centerMass[0] * -1), b + 10, a + 10, edgecolor='black', facecolor='none',
linewidth=5)
subplot.add_patch(ell)
# plt.plot([y positions of the points], [x positions of the points])
# plt.plot([testSkewLeft[1], testSkewRight[1]], [testSkewLeft[0], testSkewRight[0]], 'k')
# plt.plot([testSkewLeft[1], testSkewLeft[1]], [testSkewLeft[0], testSkewRight[0]], 'k')
# Save Scanned Signature
plt.savefig("test.PNG", dpi=700)
plt.close()
############################################
# Signature image
############################################
logoPic1 = QPixmap(filename)
self.signature.setPixmap(logoPic1.scaled(291,255,Qt.KeepAspectRatio))
self.signature.show()
############################################
# Signature Feature
############################################
search_pixmap = QtGui.QPixmap()
search_pixmap.load('test.PNG')
self.featureSignature.setPixmap(search_pixmap.scaled(291, 255, Qt.KeepAspectRatio))
self.featureSignature.show()
########################
# Save test image
##########################
vectorFeature = []
vectorFeature.append(round(centerMass[0], 4))
vectorFeature.append(round(centerMass[1], 4))
vectorFeature.append(round(eccentricity, 4))
vectorFeature.append(round(kurtosis[0], 4))
vectorFeature.append(round(kurtosis[1], 4))
vectorFeature.append(round(matrixGradient[0][0][0], 4))
vectorFeature.append(round(matrixGradient[0][0][1], 4))
vectorFeature.append(round(matrixGradient[0][1][0], 4))
vectorFeature.append(round(matrixGradient[0][1][1], 4))
vectorFeature.append(round(matrixGradient[0][2][0], 4))
vectorFeature.append(round(matrixGradient[0][2][1], 4))
vectorFeature.append(round(matrixGradient[0][3][0], 4))
vectorFeature.append(round(matrixGradient[0][3][1], 4))
vectorFeature.append(round(matrixGradient[1][0][0], 4))
vectorFeature.append(round(matrixGradient[1][0][1], 4))
vectorFeature.append(round(matrixGradient[1][1][0], 4))
vectorFeature.append(round(matrixGradient[1][1][1], 4))
vectorFeature.append(round(matrixGradient[1][2][0], 4))
vectorFeature.append(round(matrixGradient[1][2][1], 4))
vectorFeature.append(round(matrixGradient[1][3][0], 4))
vectorFeature.append(round(matrixGradient[1][3][1], 4))
vectorFeature.append(round(matrixPresure[0][0], 4))
vectorFeature.append(round(matrixPresure[0][1], 4))
vectorFeature.append(round(matrixPresure[0][2], 4))
vectorFeature.append(round(matrixPresure[0][3], 4))
vectorFeature.append(round(matrixPresure[1][0], 4))
vectorFeature.append(round(matrixPresure[1][1], 4))
vectorFeature.append(round(matrixPresure[1][2], 4))
vectorFeature.append(round(matrixPresure[1][3], 4))
vectorFeature.append(round(matrixPresure[2][0], 4))
vectorFeature.append(round(matrixPresure[2][1], 4))
vectorFeature.append(round(matrixPresure[2][2], 4))
vectorFeature.append(round(matrixPresure[2][3], 4))
vectorFeature.append(round(matrixPresure[3][0], 4))
vectorFeature.append(round(matrixPresure[3][1], 4))
vectorFeature.append(round( matrixPresure[3][2], 4))
vectorFeature.append(round(matrixPresure[3][3], 4))
label = 0
if len(filename.split('/')[len(filename.split('/'))-1]) < 11:
label = int(filename.split('/')[len(filename.split('/'))-2])
# print label
matrixData = []
vectorTarget = []
with open('../LearningInformation/learningBayesianMulticlassGenuine.csv') as f:
readerGenuine = csv.reader(f)
readerGenuine.next()
for i in readerGenuine:
matrixData.append(map(float, i[:len(i) - 1]))
# matrixData.append(map(float, i[2:5] + i[25:len(i) - 1]))
vectorTarget.append(int(i[len(i) - 1]))
with open('../LearningInformation/learningBayesianMulticlassForgeries.csv') as f:
readerForgeriestmp = csv.reader(f)
readerForgeriestmp.next()
for i in readerForgeriestmp:
matrixData.append(map(float, i[:len(i) - 1]))
# matrixData.append(map(float, i[2:5] + i[25:len(i) - 1]))
vectorTarget.append(int(i[len(i) - 1]))
matrixData = np.array(matrixData)
vectorTarget = np.array(vectorTarget)
asdsada = TemplateClassifier().fit(matrixData, vectorTarget)
y_predict_X = asdsada.predict(vectorFeature)[0]
self.realLabelText = label
self.predicted = y_predict_X