contours.py

import cv2, math
import numpy as np
from coords import *



def getContourWithArea(cnts, area, floor=500, ceil=1000):
    if cnts==None or len(cnts) == 0: return None
    # m00 moment of a contour = area of enclosed blob
    handCnt = min(cnts, key=lambda x: abs(area - cv2.moments(x)['m00']))
    if floor < cv2.moments(handCnt)['m00'] and cv2.moments(handCnt)['m00'] < ceil: return handCnt
    else: return None


def getBiggestContour(cnts):
    if cnts==None or len(cnts) == 0: return None
    # m00 moment of a contour = area of enclosed blob
    handCnt = max(cnts, key=lambda x: cv2.moments(x)['m00'])
    return handCnt


def getBottomLeftPoint(cnt):
    if cnt == None: return None
    x, y, w, h = cv2.boundingRect(cnt)
    bottomLeftPoint = Point(x, y+h)
    return bottomLeftPoint


def getBottomRightPoint(cnt):
    if cnt == None: return None
    x, y, w, h = cv2.boundingRect(cnt)
    bottomRightPoint = Point(x+w, y+h)
    return bottomRightPoint


def anyHullVerticesNear(cnt, point, radius=500):
    if cnt == None: return None
    hullPnts = getUniqueHullPoints(cnt)
    nearPnts = [p for p in hullPnts if p.getDistTo(point) < radius]
    return len(nearPnts) > 0


def getUniqueHullPoints(cnt):
    """Point is unique if not too close to any other points (>10px away from other pnts)."""
    if cnt == None: return None
    hullPnts = [Point(pnt[0][0], pnt[0][1]) for pnt in cv2.convexHull(cnt).tolist()]  # pnt = [[x,y]], pnt[0] = [x,y]
    uniquePnts = [hullPnts[i] for i in range(len(hullPnts)) if hullPnts[i].getDistTo(hullPnts[(i+1) % len(hullPnts)]) > 10]
    return uniquePnts


def getContourConvexDefects(cnt, minSize=0, maxSize=0):
    if cnt == None: return None
    hull = cv2.convexHull(cnt, returnPoints=False)
    defects = cv2.convexityDefects(cnt, hull).tolist()
    maxDefectIndices = [pnt[0][2] for pnt in defects]
    maxDefectPoints = [Point(cnt[f][0][0], cnt[f][0][1]) for f in maxDefectIndices]
    maxDefectDists = [pnt[0][3]/256 for pnt in defects]  # unknown why dist/256 gives proper value
    maxDefectPoints = filter(lambda pnt: maxDefectDists[maxDefectPoints.index(pnt)] in range(minSize, maxSize), maxDefectPoints)
    return maxDefectPoints


def getApproxContourPolygon(cnt, accuracy=0.01):
    if cnt == None: return None
    epsilon = accuracy*cv2.arcLength(cnt, True)
    approxPoly = cv2.approxPolyDP(cnt, epsilon, True).tolist()
    return [Point(pnt[0][0], pnt[0][1]) for pnt in approxPoly]


def getCntPntLs(cnt):
    return [Point(pnt[0][0], pnt[0][1]) for pnt in cnt.tolist()]


def isPointInContour(pnt, cnt):
    return cv2.pointPolygonTest(cnt, pnt.toTuple(), False) == 1