OpenCV with Python - KNN Example

KNN Ref:
http://www.hudong.com/wiki/KNN
http://en.wikipedia.org/wiki/K-nearest_neighbor_algorithm

Example Ref: (C Language)
http://www.opencv.org.cn/index.php/%E6%9C%BA%E5%99%A8%E5%AD%A6%E4%B9%A0%E4%B8%AD%E6%96%87%E5%8F%82%E8%80%83%E6%89%8B%E5%86%8C

from ml import *
from highgui import *
from cv import *

if __name__ == '__main__':
    color = CV_RGB(180, 120, 0)
    K = 10
    trainSampleCount = 300
    rngState = cvRNG()
    trainData = cvCreateMat(trainSampleCount, 2, CV_32FC1)
    trainClasses = cvCreateMat(trainSampleCount, 1, CV_32FC1)
    img = cvCreateImage(cvSize(500, 500), 8, 3)
    sample = cvCreateMat(1, 2, CV_32FC1)
    cvZero(img)

    # form the training samples
    trainData1 = cvGetRows(trainData, 0, 100)
    colData1x = cvGetCol(trainData1, 0)
    colData1y = cvGetCol(trainData1, 1)
    cvRandArr(rngState, colData1x, CV_RAND_NORMAL, cvScalar(200), cvScalar(50))
    cvRandArr(rngState, colData1y, CV_RAND_NORMAL, cvScalar(200), cvScalar(50))

    trainData2 = cvGetRows(trainData, 100, 200)
    colData2x = cvGetCol(trainData2, 0)
    colData2y = cvGetCol(trainData2, 1)
    cvRandArr(rngState, colData2x, CV_RAND_NORMAL, cvScalar(300), cvScalar(50))
    cvRandArr(rngState, colData2y, CV_RAND_NORMAL, cvScalar(300), cvScalar(50))

    trainData3 = cvGetRows(trainData, 200, 300)
    colData3x = cvGetCol(trainData3, 0)
    colData3y = cvGetCol(trainData3, 1)
    cvRandArr(rngState, colData3x, CV_RAND_NORMAL, cvScalar(100), cvScalar(30))
    cvRandArr(rngState, colData3y, CV_RAND_NORMAL, cvScalar(400), cvScalar(30))

    trainClasses1 = cvGetRows(trainClasses, 0, 100)
    cvSet(trainClasses1, cvScalar(1))
    
    trainClasses2 = cvGetRows(trainClasses, 100, 200)
    cvSet(trainClasses2, cvScalar(2))
    
    trainClasses3 = cvGetRows(trainClasses, 200, 300)
    cvSet(trainClasses3, cvScalar(3))

    knn = CvKNearest(trainData, trainClasses, None, False, K)
    nearests = cvCreateMat(1, K, CV_32FC1)

    for i in range(0, img.height):
        for j in range(0, img.width):
            sample[0, 0] = float(j)
            sample[0, 1] = float(i)
            #response = knn.find_nearest(sample, K, None, 0.0, nearests, None)
            response = knn.find_nearest(sample, K, None)
            accuracy = 0
            for k in range(0, K):
                if nearests[0, k] == response:
                    accuracy += 1
            color = CV_RGB(180, 120, 0)
            if response == 1:
                if accuracy > 5:
                    color = CV_RGB(180, 0, 0)
                else:
                    color = CV_RGB(180, 100, 100)
            elif response == 2:
                if accuracy > 5:
                    color = CV_RGB(0, 180, 0)
                else:
                    color = CV_RGB(100, 180, 100)
            elif response == 3:
                if accuracy > 5:
                    color = CV_RGB(0, 0, 180)
                else:
                    color = CV_RGB(100, 100, 180)
            cvSet2D(img, i, j, color)
    try:
        for i in range(0, 100):
            pt = CvPoint()
            pt.x = cvRound(trainData1[0, i * 2])
            pt.y = cvRound(trainData1[0, i * 2 + 1])
            cvCircle(img, pt, 2, CV_RGB(255, 0, 0), CV_FILLED)
            pt.x = cvRound(trainData2[0, i * 2])
            pt.y = cvRound(trainData2[0, i * 2 + 1])
            cvCircle(img, pt, 2, CV_RGB(0, 255, 0), CV_FILLED)
            pt.x = cvRound(trainData3[0, i * 2])
            pt.y = cvRound(trainData3[0, i * 2 + 1])
            cvCircle(img, pt, 2, CV_RGB(0, 0, 255), CV_FILLED)
    except Exception, e:
        print e
    cvNamedWindow('result', 1)
    cvShowImage('result', img)
    cvWaitKey(0)
    cvReleaseMat(trainClasses)
    cvReleaseMat(trainData)

Issue:
1. knn.find_nearest(sample, K, None, 0.0, nearests, None)
Will cause Exception:
NotImplementedError: Wrong number of arguments for overloaded function
'CvKNearest_find_nearest'.
Waiting for solution: http://tech.groups.yahoo.com/group/OpenCV/message/73337

2. 'for i in range(0, 100)' will cause out of bound error. Maybe relates with Issue 1.