摘要：算法的基本原理簡(jiǎn)單假設(shè)我們有一堆點(diǎn)，和一個(gè)小的圓形窗口，算法就是不斷移動(dòng)小圓形窗口，直到找到圓形區(qū)域內(nèi)最大灰度密度處為止初始窗口以藍(lán)色圓圈顯示，名稱為其原始中心標(biāo)有藍(lán)色矩形，名為但是，這個(gè)窗口當(dāng)中所有點(diǎn)的點(diǎn)集構(gòu)成的質(zhì)心在藍(lán)色圓形點(diǎn)處圓環(huán)的型

Meanshift and Camshift

Meanshift 算法的基本原理簡(jiǎn)單,假設(shè)我們有一堆點(diǎn)，和一個(gè)小的圓形窗口，Meanshift 算法就是不斷移動(dòng)小圓形窗口，直到找到圓形區(qū)域內(nèi)最大灰度密度處為止.

初始窗口以藍(lán)色圓圈顯示，名稱為“C1”,其原始中心標(biāo)有藍(lán)色矩形，名為“C1_o”.但是，這個(gè)窗口當(dāng)中所有點(diǎn)的點(diǎn)集構(gòu)成的質(zhì)心在藍(lán)色圓形點(diǎn)處,圓環(huán)的型心和質(zhì)心并不重合,所以,移動(dòng)藍(lán)色的窗口以使型心與之前得到的質(zhì)心重合.
不斷執(zhí)行上面的移動(dòng)過程，直到型心和質(zhì)心大致重合結(jié)束.
通常通過直方圖反投影圖像和初始目標(biāo)位置,當(dāng)物體移動(dòng)時(shí),移動(dòng)反映在直方圖反投影圖像中,最后圓形的窗口會(huì)落到像素分布最大的地方，也就是圖中的綠色圈并命名為C2.

首先要設(shè)定目標(biāo)，并計(jì)算的直方圖，然后對(duì)這個(gè)直方圖在每一幀當(dāng)中進(jìn)行反向投影.需要提供一個(gè)初試的窗口位置,計(jì)算HSV模型當(dāng)中H(色調(diào))的直方圖,為了避免低亮度造成的影響，使用 cv2.inRange()將低亮度值忽略.

import numpy as np
import cv2
import matplotlib.pyplot as plt

cap = cv2.VideoCapture("test.mp4")
# take first frame of the video
ret,frame = cap.read()

# setup initial location of window
r,h,c,w = 50,200,50,100  # simply hardcoded the values
track_window = (c,r,w,h)

# set up the ROI for tracking
roi = frame[r:r+h, c:c+w]
hsv_roi =  cv2.cvtColor(roi, cv2.COLOR_BGR2HSV)
mask = cv2.inRange(hsv_roi, np.array((0., 60.,32.)), np.array((180.,255.,255.)))
roi_hist = cv2.calcHist([hsv_roi],[0],mask,[180],[0,180])
cv2.normalize(roi_hist,roi_hist,0,255,cv2.NORM_MINMAX)

# Setup the termination criteria, either 10 iteration or move by atleast 1 pt
term_crit = ( cv2.TERM_CRITERIA_EPS | cv2.TERM_CRITERIA_COUNT, 10, 1 )

while(1):
    ret ,frame = cap.read()

    if ret == True:
        hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)
        dst = cv2.calcBackProject([hsv],[0],roi_hist,[0,180],1)

        # apply meanshift to get the new location
        ret, track_window = cv2.CamShift(dst, track_window, term_crit)

        # Draw it on image
        pts = cv2.boxPoints(ret)
        pts = np.int0(pts)
        img2 = cv2.polylines(frame,[pts],True, 255,2)
        cv2.imshow("img2",img2)

        k = cv2.waitKey(60) & 0xff
        if k == 27:
            break
        else:
            cv2.imwrite(chr(k)+".jpg",img2)

    else:
        break

cv2.destroyAllWindows()
cap.release()

在目標(biāo)跟蹤中，物體的大小不是固定的，所以設(shè)置的跟蹤窗口也應(yīng)該隨之變化，CAMshift算法，首先使用meanshift算法找到目標(biāo)，然后調(diào)整窗口大小，而且還會(huì)計(jì)算目標(biāo)對(duì)象的的最佳外接圓的角度，并調(diào)整窗口，并使用調(diào)整后的窗口對(duì)物體繼續(xù)追蹤.

它與meanshift幾乎相同，但它返回一個(gè)旋轉(zhuǎn)的矩形.

import numpy as np
import cv2
import matplotlib.pyplot as plt

cap = cv2.VideoCapture("test.mp4")
# take first frame of the video
ret,frame = cap.read()

# setup initial location of window
r,h,c,w = 50,200,50,100  # simply hardcoded the values
track_window = (c,r,w,h)

# set up the ROI for tracking
roi = frame[r:r+h, c:c+w]
hsv_roi =  cv2.cvtColor(roi, cv2.COLOR_BGR2HSV)
mask = cv2.inRange(hsv_roi, np.array((0., 60.,32.)), np.array((180.,255.,255.)))
roi_hist = cv2.calcHist([hsv_roi],[0],mask,[180],[0,180])
cv2.normalize(roi_hist,roi_hist,0,255,cv2.NORM_MINMAX)

# Setup the termination criteria, either 10 iteration or move by atleast 1 pt
term_crit = ( cv2.TERM_CRITERIA_EPS | cv2.TERM_CRITERIA_COUNT, 10, 1 )

while(1):
    ret ,frame = cap.read()

    if ret == True:
        hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)
        dst = cv2.calcBackProject([hsv],[0],roi_hist,[0,180],1)

        # apply meanshift to get the new location
        ret, track_window = cv2.CamShift(dst, track_window, term_crit)

        # Draw it on image
        pts = cv2.boxPoints(ret)
        pts = np.int0(pts)
        img2 = cv2.polylines(frame,[pts],True, 255,2)
        cv2.imshow("img2",img2)

        k = cv2.waitKey(60) & 0xff
        if k == 27:
            break
        else:
            cv2.imwrite(chr(k)+".jpg",img2)

    else:
        break

cv2.destroyAllWindows()
cap.release()