computer vision 4 5

.py

#!/usr/bin/python
# -*- coding: utf-8 -*-
import cv2
import numpy as np
from sklearn.naive_bayes import GaussianNB
from sklearn.neighbors import KNeighborsClassifier
from sklearn.neural_network import MLPClassifier

shapeData = list()
label = [4,3,2,2,3,3,4,1,2,1,1,1,1,4,4,1,3,1,2,1,1,3,3,4]

def calculateEccentricity(M):
    xhat = M['m10'] / M['m00']
    yhat = M['m01'] / M['m00']

    u20 = M['m20'] / M['m00'] - xhat ** 2
    u02 = M['m02'] / M['m00'] - yhat ** 2
    u11 = M['m11'] / M['m00'] - xhat * yhat

    e1 = (u20 + u02) / 2 + (4 * u11 ** 2 + (u20 - u02) ** 2) ** (1 / 2) / 0.5
    e2 = (u20 + u02) / 2 - (4 * u11 ** 2 + (u20 - u02) ** 2) ** (1 / 2) / 0.5

    res = (1 - e2 / e1) ** 0.5

    return res

def extractFeatures(clf):
    testImg = cv2.imread("test.jpg")
    gray = cv2.cvtColor(testImg, cv2.COLOR_BGR2GRAY)
    ret, thresh = cv2.threshold(gray, 0, 127, cv2.THRESH_OTSU)
    (im2, contours, hierarcy) = cv2.findContours(thresh, cv2.RETR_TREE,
            cv2.CHAIN_APPROX_SIMPLE)
    i = 0
    for cnt in contours:
        if i == 0:
            i += 1
            continue
    
        # put text position finding
        pos = sum(cnt) / len(cnt)
        pos = pos[0]
        x = int(pos[0]) - 15
        y = int(pos[1])
    
        # hull convexity - region based feature 1
    
        hull = cv2.convexHull(cnt)
    
        # feature moments - region based feature 2
    
        M = cv2.moments(cnt)
        m0 = M['m00']
        cx = int(M['m10'] / M['m00'])
        cy = int(M['m01'] / M['m00'])
        cx = int(M['mu20'] / M['mu02'])
        cy = int(M['m01'] / M['m00'])
    
        # corner approximation - contour based feature 1
    
        peri = cv2.arcLength(cnt, True)
        approx = cv2.approxPolyDP(cnt, 0.04 * peri, True)
    
        eccentricity = calculateEccentricity(M)
        aday = np.array([[len(approx),len(hull)/peri,eccentricity]])
        sonuc = clf.predict(aday)
        sonuc = str(sonuc[0])
        if sonuc == "1":
            sonuc = "kare"
        elif sonuc == "2":
            sonuc = "ucgen"
        elif sonuc == "3":
            sonuc = "cember"
        elif sonuc == "4":
            sonuc = "elips"
        cv2.putText(testImg,
            str(sonuc),
            (x, y),
            cv2.FONT_HERSHEY_SIMPLEX,
            0.7,
            (0, 0, 255),
            1,)
    
    return testImg

def bayesClf():
    global shapeData
    clf = GaussianNB()
    shapeData = np.array(shapeData)
    clf.fit(shapeData, label)

    cv2.imshow("bayes",extractFeatures(clf))

def knnClf():
    global shapeData
    clf = KNeighborsClassifier(n_neighbors=3)
    clf.fit(shapeData, label)

    cv2.imshow("knn",extractFeatures(clf))

def neuralClf():
    global shapeData
    clf = MLPClassifier()
    clf.fit(shapeData, label)

    cv2.imshow("neural", extractFeatures(clf))

def main():
    org = cv2.imread('shapes.jpg')
    gray = cv2.cvtColor(org, cv2.COLOR_BGR2GRAY)
    (ret, thresh) = cv2.threshold(gray, 0, 127, cv2.THRESH_OTSU)

    (im2, contours, hierarcy) = cv2.findContours(thresh, cv2.RETR_TREE,
            cv2.CHAIN_APPROX_SIMPLE)
    cv2.drawContours(org, contours, -1, (255, 25, 0), 2)

    i = 0
    for cnt in contours:
        if i == 0:
            i += 1
            continue

        # put text position finding

        pos = sum(cnt) / len(cnt)
        pos = pos[0]
        x = int(pos[0]) - 15
        y = int(pos[1])

        # hull convexity - region based feature 1

        hull = cv2.convexHull(cnt)

        # feature moments - region based feature 2

        M = cv2.moments(cnt)
        m0 = M['m00']
        cx = int(M['m10'] / M['m00'])
        cy = int(M['m01'] / M['m00'])
        cx = int(M['mu20'] / M['mu02'])
        cy = int(M['m01'] / M['m00'])

        # corner approximation - contour based feature 1

        peri = cv2.arcLength(cnt, True)
        approx = cv2.approxPolyDP(cnt, 0.04 * peri, True)

        # eccentricity

        eccentricity = calculateEccentricity(M)

        cv2.putText(org,
            'i : ' + str(i),
            (x, y),
            cv2.FONT_HERSHEY_SIMPLEX,
            0.4,
            (0, 0, 255),
            1,)
        cv2.putText(org,
            's : ' + str(len(approx)),
            (x, y + 15),
            cv2.FONT_HERSHEY_SIMPLEX,
            0.4,
            (0, 0, 255),
            1,)
        cv2.putText(org,
            'h : ' + str('{:.2f}'.format(len(hull) / peri)),
            (x, y + 30),
            cv2.FONT_HERSHEY_SIMPLEX,
            0.4,
            (0, 0, 255),
            1,)
        cv2.putText(org,
            'e: ' + str('{:.2f}'.format(eccentricity)),
            (x, y - 15),
            cv2.FONT_HERSHEY_SIMPLEX,
            0.4,
            (0, 0, 255),
            1,)
        i += 1
        shapeData.append([len(approx),len(hull)/peri,eccentricity])
        print("[" + str(i) + "," +  str(len(approx)) + "," + str('{:.2f}'.format(len(hull) / peri)) + "," +str('{:.2f}'.format(eccentricity)) + "],",sep='')
    
    bayesClf()
    knnClf()
    neuralClf()

    cv2.imshow('shapes', org)
    cv2.waitKey(0)
    cv2.destroyAllWindows()


if __name__ == '__main__':
    main()