一、前言

首先要知道，通过openpose对手势进行识别，需要找到骨骼点，才能确定手部和脸部的位置，再识别手部和脸部。

前一篇文章讲了对 骨骼关键点 进行划分，再找到手部感兴趣图像（ROI）：
Openpose人体骨骼、手势–静态图像标记及分类（附源码）

本文是对 手势特征 进行划分，
根据2008年，北京师范大学的特殊教育研究学者洛维维对《中国手语》进行归纳统计分析后，将5k多个中国手语词汇的基本手型归纳为61个（见后表）。

思路：

对于每幅图像提取出手型：指尖和重心（这里以手腕为参考点非实际重心），然后计算出距离和夹角，对于不同手势分别进行距离和夹角的统计，得到其分布的数字特征。

参考于：《基于计算机视觉的手势识别研究》

二、手势

1.分析

这是简单的手语拼音的部分采集图像：

根据openpose【22关键点】识别手语效果如下：

可以发现，主要是手指头的判别,同coco骨骼模型判别，我们将手势特征划分为：距离和角度。

通过手指到手腕的距离和手指的角度，我们能很好的得到手势的关键信息！
这里划分标准为：【21个手势点，第22个点为背景】

当然这种划分情况，在识别不到手腕Wrist点时，会失真，这里暂不考虑这种特殊情况，后面都以假设0点存在进行讨论。

关键函数：

    def __distance(self,A,B):"""距离辅助函数:param 两个坐标A(x1,y1)B(x2,y2):return 距离d=AB的距离"""if A is None or B is None:return 0else:return math.sqrt((A[0]-B[0])**2+(A[1]-B[1])**2)def __myAngle(self,A,B,C):"""角度辅助函数:param 三个坐标A(x1,y1)B(x2,y2)C(x3,y3):return 角B的余弦值（转换为角度）"""if A is None or B is None or C is None:return 0else:a=self.__distance(B,C)b=self.__distance(A,C)c=self.__distance(A,B)if 2*a*c !=0:return math.degrees(a**2/+c**2-b**2)/(2*a*c)#计算出cos弧度，转换为角度return 0def handDistance(self,rkeyPoint,lkeyPoint):"""距离辅助函数:param keyPoint::return:list:distance:""" if keyPoint[0] is None:print("未识别到Wrist参考关键点")distance0 = self.__distance(keyPoint[0],keyPoint[4])#Thumb拇指distance1 = self.__distance(keyPoint[0],keyPoint[8])#Index食指distance2 = self.__distance(keyPoint[0],keyPoint[12])#Middle中指distance3 = self.__distance(keyPoint[0],keyPoint[16])#Ring无名指distance4 = self.__distance(keyPoint[0],keyPoint[20])#Little小指return [distance0, distance1, distance2, distance3, distance4]def handpointAngle(self, keyPoint):"""角度辅助函数:param keyPoint::return:list:角度:"""angle0 = self.__myAngle(keyPoint[0], keyPoint[9], keyPoint[4])angle1 = self.__myAngle(keyPoint[0], keyPoint[9], keyPoint[8])angle2 = self.__myAngle(keyPoint[0], keyPoint[9], keyPoint[12])angle3 = self.__myAngle(keyPoint[0], keyPoint[9], keyPoint[16])angle4 = self.__myAngle(keyPoint[0], keyPoint[9], keyPoint[20])return [angle0, angle1, angle2, angle3, angle4]def getHandsInformation(self,rpoints,lpoints):"""将右手和左手（各距离5和角度5个特征）信息汇集:param 左右手关键点:return 特征集合共20特征点"""Information = []#右手DistanceList = pose_model.bonepointDistance(rpoints)#  距离关键信息AngleList = pose_model.bonepointAngle(rpoints)#  角度关键信息for i in range(len(DistanceList)):Information.append(DistanceList[i])for j in range(len(AngleList)):Information.append(AngleList[j])#左手DistanceList = pose_model.bonepointDistance(lpoints)#  距离关键信息AngleList = pose_model.bonepointAngle(lpoints)#  角度关键信息for m in range(len(DistanceList)):Information.append(DistanceList[m])for n in range(len(AngleList)):Information.append(AngleList[n])return Information

2.效果演示

标注

提取

距离和角度计算

定位

标注

距离和角度计算

图像

骨骼检测

骨骼图像

骨骼特征点

有效信息

手部图像

手势检测

手部特征点

3.完整代码

后半部分是傅里叶描述子呈现肤色提取，这里不适用！要求手语者身着长袖，且背景不与肤色相近。据情况选择。

参考图片：（猜猜是什么意思？正确答案是：梨）

源码：

#!/usr/bin/python3
#!--*-- coding: utf-8 --*--
from __future__ import division# 精确除法
import cv2
import os
import time
import math
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
plt.rcParams['font.sans-serif']=['SimHei'] #用来正常显示中文标签
plt.rcParams['axes.unicode_minus']=False #用来正常显示负号class general_pose_model(object):def __init__(self, modelpath):# 指定采用的模型#   hand: 22 points（21个手势关键点，第22个点表示背景）#   COCO:   18 points（）self.inWidth = 368self.inHeight = 368self.threshold = 0.1self.pose_net = self.general_coco_model(modelpath)self.hand_num_points = 22self.hand_point_pairs = [[0,1],[1,2],[2,3],[3,4],[0,5],[5,6],[6,7],[7,8],[0,9],[9,10],[10,11],[11,12],[0,13],[13,14],[14,15],[15,16],[0,17],[17,18],[18,19],[19,20]]self.hand_net = self.get_hand_model(modelpath)self.MIN_DESCRIPTOR = 32  # surprisingly enough, 2 descriptors are already enough"""提取骨骼特征点，并可视化显示"""def general_coco_model(self, modelpath):"""COCO输出格式：鼻子-0, 脖子-1，右肩-2，右肘-3，右手腕-4，左肩-5，左肘-6，左手腕-7，右臀-8，右膝盖-9，右脚踝-10，左臀-11，左膝盖-12，左脚踝-13，右眼-14，左眼-15，有耳朵-16，左耳朵-17，背景-18."""self.points_name = {"Nose": 0, "Neck": 1, "RShoulder": 2, "RElbow": 3, "RWrist": 4,"LShoulder": 5, "LElbow": 6, "LWrist": 7, "RHip": 8, "RKnee": 9, "RAnkle": 10, "LHip": 11, "LKnee": 12, "LAnkle": 13, "REye": 14, "LEye": 15, "REar": 16, "LEar": 17, "Background": 18}self.bone_num_points = 18self.bone_point_pairs = [[1, 0], [1, 2], [1, 5], [2, 3], [3, 4], [5, 6], [6, 7], [1, 8], [8, 9],[9, 10], [1, 11], [11, 12], [12, 13], [0, 14], [0, 15], [14, 16], [15, 17]]prototxt   = os.path.join(modelpath,"pose/coco/pose_deploy_linevec.prototxt")caffemodel = os.path.join(modelpath, "pose/coco/pose_iter_440000.caffemodel")coco_model = cv2.dnn.readNetFromCaffe(prototxt, caffemodel)return coco_modeldef getBoneKeypoints(self, imgfile):"""COCO身体关键点检测:param 图像路径:return 关键点坐标集合"""img_cv2 = cv2.imread(imgfile)img_height, img_width, _ = img_cv2.shape#读取图像并生成输入blobinpBlob = cv2.dnn.blobFromImage(img_cv2,1.0 / 255,(self.inWidth, self.inHeight),(0, 0, 0), swapRB=False, crop=False)#向前通过网络self.pose_net.setInput(inpBlob)self.pose_net.setPreferableBackend(cv2.dnn.DNN_BACKEND_OPENCV)self.pose_net.setPreferableTarget(cv2.dnn.DNN_TARGET_OPENCL)output = self.pose_net.forward()H = output.shape[2]W = output.shape[3]print("形状：")print(output.shape)# vis heatmapsself.vis_bone_heatmaps(img_file, output)#points = []for idx in range(self.bone_num_points):#把输出的大小调整到与输入一样probMap = output[0, idx, :, :] # confidence map.# 提取关键点区域的局部最大值minVal, prob, minLoc, point = cv2.minMaxLoc(probMap)# Scale the point to fit on the original imagex = (img_width * point[0]) / Wy = (img_height * point[1]) / Hif prob > self.threshold:points.append((int(x), int(y)))else:points.append(None)#print(points)return pointsdef __distance(self,A,B):"""距离辅助函数:param 两个坐标A(x1,y1)B(x2,y2):return 距离d=AB的距离"""if A is None or B is None:return 0else:return math.sqrt((A[0]-B[0])**2+(A[1]-B[1])**2)def __myAngle(self,A,B,C):"""角度辅助函数:param 三个坐标A(x1,y1)B(x2,y2)C(x3,y3):return 角B的余弦值（转换为角度）"""if A is None or B is None or C is None:return 0else:a=self.__distance(B,C)b=self.__distance(A,C)c=self.__distance(A,B)if 2*a*c !=0:return math.degrees(a**2/+c**2-b**2)/(2*a*c)#计算出cos弧度，转换为角度return 0def bonepointDistance(self, keyPoint):"""距离辅助函数:param keyPoint::return:list:distance:"""distance0 = self.__distance(keyPoint[4],keyPoint[8])#右手右腰distance1 = self.__distance(keyPoint[7],keyPoint[11])#左手左腰distance2 = self.__distance(keyPoint[2],keyPoint[4])#手肩distance3 = self.__distance(keyPoint[5],keyPoint[7])distance4 = self.__distance(keyPoint[0],keyPoint[4])#头手distance5 = self.__distance(keyPoint[0],keyPoint[7])distance6 = self.__distance(keyPoint[4],keyPoint[7])#两手distance7 = self.__distance(keyPoint[4],keyPoint[16])#手耳distance8 = self.__distance(keyPoint[7],keyPoint[17])distance9 = self.__distance(keyPoint[4],keyPoint[14])#手眼distance10 = self.__distance(keyPoint[7],keyPoint[15])distance11 = self.__distance(keyPoint[4],keyPoint[1])#手脖distance12 = self.__distance(keyPoint[7],keyPoint[1])distance13 = self.__distance(keyPoint[4],keyPoint[5])#左手左臂distance14 = self.__distance(keyPoint[4],keyPoint[6])#右手左肩distance15 = self.__distance(keyPoint[7],keyPoint[2])#右手左肩distance16 = self.__distance(keyPoint[7],keyPoint[3])#左手右臂return [distance0, distance1, distance2, distance3, distance4, distance5, distance6, distance7,distance8,distance9, distance10, distance11, distance12, distance13, distance14, distance15, distance16]def bonepointAngle(self, keyPoint):"""角度辅助函数:param keyPoint::return:list:角度:"""angle0 = self.__myAngle(keyPoint[2], keyPoint[3], keyPoint[4])#右手臂夹角angle1 = self.__myAngle(keyPoint[5], keyPoint[6], keyPoint[7])#左手臂夹角angle2 = self.__myAngle(keyPoint[3], keyPoint[2], keyPoint[1])#右肩夹角angle3 = self.__myAngle(keyPoint[6], keyPoint[5], keyPoint[1])angle4 = self.__myAngle(keyPoint[4], keyPoint[0], keyPoint[7])#头手头if keyPoint[8] is None or keyPoint[11] is None:angle5 = 0else:temp = ((keyPoint[8][0]+keyPoint[11][0])/2,(keyPoint[8][1]+keyPoint[11][1])/2)#两腰的中间值angle5 = self.__myAngle(keyPoint[4], temp, keyPoint[7])#手腰手angle6 = self.__myAngle(keyPoint[4], keyPoint[1], keyPoint[8])#右手脖腰angle7 = self.__myAngle(keyPoint[7], keyPoint[1], keyPoint[11])#右手脖腰return [angle0, angle1, angle2, angle3, angle4, angle5, angle6, angle7]def getBoneInformation(self,bone_points):"""将距离和角度25个特征信息汇集:param 骨骼关键点:return list 距离和角度25个特征信息"""Information = []#print("骨骼关键距离信息: ")DistanceList = self.bonepointDistance(bone_points)# 3. 距离关键信息#print(DistanceList)#print("骨骼关键角度信息: ")AngleList = self.bonepointAngle(bone_points)# 4. 角度关键信息#print(AngleList)for i in range(len(DistanceList)):Information.append(DistanceList[i])for j in range(len(AngleList)):Information.append(AngleList[j])return Informationdef vis_bone_pose(self,imgfile,points):"""显示标注骨骼点后的图像:param 图像路径，COCO检测关键点坐标:return 骨骼连线图、关键点图"""img_cv2 = cv2.imread(imgfile)img_cv2_copy = np.copy(img_cv2)for idx in range(len(points)):if points[idx]:cv2.circle(img_cv2_copy, points[idx], 5, (0, 255, 255), thickness=-1,lineType=cv2.FILLED)cv2.putText(img_cv2_copy, "{}".format(idx), points[idx], cv2.FONT_HERSHEY_SIMPLEX,1,(0, 0, 255),4, lineType=cv2.LINE_AA)h = int(self.__distance(points[4],points[3]))#小臂周长if points[4]:x_center = points[4][0]y_center = points[4][1]cv2.rectangle(img_cv2_copy, (x_center-h, y_center-h), (x_center+h, y_center+h), (255, 0, 0), 2)#框cv2.circle(img_cv2_copy,(x_center, y_center), 1, (100, 100, 0), thickness=-1,lineType=cv2.FILLED)#坐标点cv2.putText(img_cv2_copy,"%d,%d" % (x_center,y_center),(x_center, y_center), cv2.FONT_HERSHEY_SIMPLEX,0.6, (100, 100, 0), 2, lineType=cv2.LINE_AA)#右手首if points[7]:x_center = points[7][0]y_center = points[7][1]cv2.rectangle(img_cv2_copy, (x_center-h, y_center-h), (x_center+h, y_center+h), (255, 0, 0), 1)cv2.putText(img_cv2_copy,"%d,%d" % (x_center,y_center),(x_center, y_center), cv2.FONT_HERSHEY_SIMPLEX,0.6, (100, 100, 0), 2, lineType=cv2.LINE_AA)#左手首cv2.circle(img_cv2_copy,(x_center-h, y_center-h), 3, (225, 225, 255), thickness=-1,lineType=cv2.FILLED)#对角点cv2.putText(img_cv2_copy, "{}".format(x_center-h),(x_center-h, y_center-h), cv2.FONT_HERSHEY_SIMPLEX,0.6, (100, 100, 0), 2, lineType=cv2.LINE_AA)cv2.circle(img_cv2_copy,(x_center+h, y_center+h), 3, (225, 225, 255), thickness=-1,lineType=cv2.FILLED)cv2.putText(img_cv2_copy, "{}".format(x_center+h),(x_center+h, y_center+h), cv2.FONT_HERSHEY_SIMPLEX,0.6, (100, 100, 0), 2, lineType=cv2.LINE_AA)#对角点# 骨骼连线for pair in self.bone_point_pairs:partA = pair[0]partB = pair[1]if points[partA] and points[partB]:cv2.line(img_cv2, points[partA], points[partB], (0, 255, 255), 3)cv2.circle(img_cv2, points[partA],4, (0, 0, 255),thickness=-1, lineType=cv2.FILLED)plt.figure(figsize=[10, 10])plt.subplot(1, 2, 1)plt.imshow(cv2.cvtColor(img_cv2, cv2.COLOR_BGR2RGB))plt.axis("off")plt.subplot(1, 2, 2)plt.imshow(cv2.cvtColor(img_cv2_copy, cv2.COLOR_BGR2RGB))plt.axis("off")plt.show()def vis_bone_heatmaps(self, imgfile, net_outputs):"""显示骨骼关键点热力图:param 图像路径，神经网络"""img_cv2 = cv2.imread(imgfile)plt.figure(figsize=[10, 10])for pdx in range(self.bone_num_points):probMap = net_outputs[0, pdx, :, :]#全部heatmap都初始化为0probMap = cv2.resize(probMap,(img_cv2.shape[1], img_cv2.shape[0]))plt.subplot(5, 5, pdx+1)plt.imshow(cv2.cvtColor(img_cv2, cv2.COLOR_BGR2RGB))# backgroundplt.imshow(probMap, alpha=0.6)plt.colorbar()plt.axis("off")plt.show()"""提取手势图像（在骨骼基础上定位左右手图片），handpose特征点，并可视化显示"""def get_hand_model(self, modelpath):prototxt   = os.path.join(modelpath, "hand/pose_deploy.prototxt")caffemodel = os.path.join(modelpath, "hand/pose_iter_102000.caffemodel")hand_model = cv2.dnn.readNetFromCaffe(prototxt, caffemodel)return hand_modeldef getOneHandKeypoints(self, handimg):"""hand手部关键点检测（单手）:param 手部图像路径，手部关键点:return 单手关键点坐标集合"""img_height, img_width, _ = handimg.shape        aspect_ratio = img_width / img_heightinWidth = int(((aspect_ratio * self.inHeight) * 8) // 8)inpBlob = cv2.dnn.blobFromImage(handimg, 1.0 / 255, (inWidth, self.inHeight), (0, 0, 0), swapRB=False, crop=False)self.hand_net.setInput(inpBlob)output = self.hand_net.forward()# vis heatmapsself.vis_hand_heatmaps(handimg, output)#points = []for idx in range(self.hand_num_points):probMap = output[0, idx, :, :] # confidence map.probMap = cv2.resize(probMap, (img_width, img_height))# Find global maxima of the probMap.minVal, prob, minLoc, point = cv2.minMaxLoc(probMap)if prob > self.threshold:points.append((int(point[0]), int(point[1])))else:points.append(None)return pointsdef getHandROI(self,imgfile,bonepoints):"""hand手部感兴趣的区域寻找到双手图像:param 图像路径，骨骼关键点:return 左手关键点，右手关键点坐标集合,原始图片位置参数"""img_cv2 = cv2.imread(imgfile)#原图像img_height, img_width, _ = img_cv2.shaperimg = img_cv2.copy()#图像备份limg = img_cv2.copy()# 以右手首为中心，裁剪长度为小臂长的图片if bonepoints[4] and bonepoints[3]:#右手h = int(self.__distance(bonepoints[4],bonepoints[3]))#小臂长x_center = bonepoints[4][0]y_center = bonepoints[4][1]x1 = x_center-hy1 = y_center-hx2 = x_center+hy2 = y_center+hprint(x1,x2,x_center,y_center,y1,y2)if x1< 0:x1 = 0if x2>img_width:x2 = img_widthif y1< 0:y1 = 0if y2>img_height:y2 = img_heightrimg = img_cv2[y1:y2,x1:x2]# 裁剪坐标为[y0:y1, x0:x1]if bonepoints[7] and bonepoints[6]:#左手h = int(self.__distance(bonepoints[7],bonepoints[6]))#小臂长x_center = bonepoints[7][0]y_center = bonepoints[7][1]x1 = x_center-hy1 = y_center-hx2 = x_center+hy2 = y_center+hprint(x1,x2,x_center,y_center,y1,y2)if x1< 0:x1 = 0if x2>img_width:x2 = img_widthif y1< 0:y1 = 0if y2>img_height:y2 = img_heightlimg = img_cv2[y1:y2,x1:x2]# 裁剪坐标为[y0:y1, x0:x1]  plt.figure(figsize=[10, 10])plt.subplot(1, 2, 1)plt.imshow(cv2.cvtColor(rimg, cv2.COLOR_BGR2RGB))plt.axis("off")plt.subplot(1, 2, 2)plt.imshow(cv2.cvtColor(limg, cv2.COLOR_BGR2RGB))plt.axis("off")plt.show()return rimg,limgdef getHandsKeypoints(self,rimg,limg):"""双手图像分别获取特征点:param 图像路径，骨骼关键点:return 左手关键点，右手关键点坐标集合"""# 分别获取手部特征点rhandpoints = self.getOneHandKeypoints(rimg)lhandpoints = self.getOneHandKeypoints(limg)#显示pose_model.vis_hand_pose(rimg, rhandpoints)pose_model.vis_hand_pose(limg, lhandpoints)return rhandpoints,lhandpointsdef handDistance(self,rkeyPoint,lkeyPoint):"""距离辅助函数:param keyPoint::return:list:distance:""" if keyPoint[0] is None:print("未识别到Wrist参考关键点")distance0 = self.__distance(keyPoint[0],keyPoint[4])#Thumb拇指distance1 = self.__distance(keyPoint[0],keyPoint[8])#Index食指distance2 = self.__distance(keyPoint[0],keyPoint[12])#Middle中指distance3 = self.__distance(keyPoint[0],keyPoint[16])#Ring无名指distance4 = self.__distance(keyPoint[0],keyPoint[20])#Little小指return [distance0, distance1, distance2, distance3, distance4]def handpointAngle(self, keyPoint):"""角度辅助函数:param keyPoint::return:list:角度:"""angle0 = self.__myAngle(keyPoint[0], keyPoint[9], keyPoint[4])angle1 = self.__myAngle(keyPoint[0], keyPoint[9], keyPoint[8])angle2 = self.__myAngle(keyPoint[0], keyPoint[9], keyPoint[12])angle3 = self.__myAngle(keyPoint[0], keyPoint[9], keyPoint[16])angle4 = self.__myAngle(keyPoint[0], keyPoint[9], keyPoint[20])return [angle0, angle1, angle2, angle3, angle4]def getHandsInformation(self,rpoints,lpoints):"""将右手和左手（各距离5和角度5个特征）信息汇集:param 左右手关键点:return 特征集合共20特征点"""Information = []#右手DistanceList = self.bonepointDistance(rpoints)#  距离关键信息AngleList = self.bonepointAngle(rpoints)#  角度关键信息for i in range(len(DistanceList)):Information.append(DistanceList[i])for j in range(len(AngleList)):Information.append(AngleList[j])#左手DistanceList = self.bonepointDistance(lpoints)#  距离关键信息AngleList = self.bonepointAngle(lpoints)#  角度关键信息for m in range(len(DistanceList)):Information.append(DistanceList[m])for n in range(len(AngleList)):Information.append(AngleList[n])return Informationdef vis_hand_heatmaps(self, handimg, net_outputs):"""显示手势关键点热力图（单手）:param 图像路径，神经网络"""plt.figure(figsize=[10, 10])for pdx in range(self.hand_num_points):probMap = net_outputs[0, pdx, :, :]probMap = cv2.resize(probMap, (handimg.shape[1], handimg.shape[0]))plt.subplot(5, 5, pdx+1)plt.imshow(cv2.cvtColor(handimg, cv2.COLOR_BGR2RGB))plt.imshow(probMap, alpha=0.6)plt.colorbar()plt.axis("off")plt.show()def vis_hand_pose(self,handimg, points):"""显示标注手势关键点后的图像（单手）:param 图像路径，每只手检测关键点坐标:return 关键点连线图，关键点图"""img_cv2_copy = np.copy(handimg)for idx in range(len(points)):if points[idx]:cv2.circle(img_cv2_copy, points[idx], 2, (0, 255, 255), thickness=-1,lineType=cv2.FILLED)cv2.putText(img_cv2_copy, "{}".format(idx), points[idx], cv2.FONT_HERSHEY_SIMPLEX,0.3,(0, 0, 255), 1, lineType=cv2.LINE_AA)# Draw Skeletonfor pair in self.hand_point_pairs:partA = pair[0]partB = pair[1]if points[partA] and points[partB]:cv2.line(handimg, points[partA], points[partB], (0, 255, 255), 2)cv2.circle(handimg, points[partA], 2, (0, 0, 255), thickness=-1, lineType=cv2.FILLED)plt.figure(figsize=[10, 10])plt.subplot(1, 2, 1)plt.imshow(cv2.cvtColor(handimg, cv2.COLOR_BGR2RGB))plt.axis("off")plt.subplot(1, 2, 2)plt.imshow(cv2.cvtColor(img_cv2_copy, cv2.COLOR_BGR2RGB))plt.axis("off")plt.show()"""提取手势傅里叶描述子"""def find_contours(self,Laplacian):"""获取连通域:param: 输入Laplacian算子（空间锐化滤波） :return: 最大连通域"""#binaryimg = cv2.Canny(res, 50, 200) #二值化，canny检测h = cv2.findContours(Laplacian,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_NONE) #寻找轮廓contour = h[0]contour = sorted(contour, key = cv2.contourArea, reverse=True)#对一系列轮廓点坐标按它们围成的区域面积进行排序return contourdef skinMask(self,roi):"""YCrCb颜色空间的Cr分量+Otsu法阈值分割算法:param res: 输入原图像:return: 肤色滤波后图像"""YCrCb = cv2.cvtColor(roi, cv2.COLOR_BGR2YCR_CB) #转换至YCrCb空间(y,cr,cb) = cv2.split(YCrCb) #拆分出Y,Cr,Cb值cr1 = cv2.GaussianBlur(cr, (5,5), 0)_, skin = cv2.threshold(cr1, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU) #Ostu处理res = cv2.bitwise_and(roi,roi, mask = skin)plt.figure(figsize=(10,10))plt.subplot(1,2,1)plt.imshow(cv2.cvtColor(roi, cv2.COLOR_BGR2RGB))plt.xlabel(u'原图',fontsize=20)plt.subplot(1,2,2)plt.imshow(cv2.cvtColor(res, cv2.COLOR_BGR2RGB))plt.xlabel(u'肤色滤波后的图像',fontsize=20)plt.show()plt.figure(figsize=(10,4))plt.subplot(1, 3, 1)hist1 = cv2.calcHist([roi], [0], None, [256], [0, 256])#直方图opencvplt.xlabel(u'opencv直方图',fontsize=20)plt.plot(hist1)plt.subplot(1, 3, 2)hist2 = np.bincount(roi.ravel(), minlength=256) #np直方图hist2, bins = np.histogram(roi.ravel(), 256, [0, 256])#np直方图ravel()二维变一维plt.plot(hist2)plt.xlabel(u'np直方图',fontsize=20)plt.subplot(1, 3, 3)plt.hist(roi.ravel(), 256, [0, 256])#matlab自带直方图plt.xlabel(u'matlab直方图',fontsize=20)plt.show()#     gray= cv2.cvtColor(roi,cv2.IMREAD_GRAYSCALE)#     equ = cv2.equalizeHist(gray)#     cv2.imshow('equalization', np.hstack((roi, equ)))  # 并排显示#     cv2.waitKey(0)# 自适应均衡化，参数可选#     plt.figure()#     clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))#     cl1 = clahe.apply(roi)#     plt.show()return resdef truncate_descriptor(self,fourier_result):"""截短傅里叶描述子:param res: 输入傅里叶描述子:return: 截短傅里叶描述子"""descriptors_in_use = np.fft.fftshift(fourier_result)#取中间的MIN_DESCRIPTOR项描述子center_index = int(len(descriptors_in_use) / 2)low, high = center_index - int(self.MIN_DESCRIPTOR / 2), center_index + int(self.MIN_DESCRIPTOR / 2)descriptors_in_use = descriptors_in_use[low:high]descriptors_in_use = np.fft.ifftshift(descriptors_in_use)return descriptors_in_usedef fourierDesciptor(self,res):"""计算傅里叶描述子:param res: 输入图片:return: 图像，描述子点"""#Laplacian算子进行八邻域检测gray = cv2.cvtColor(res, cv2.COLOR_BGR2GRAY)dst = cv2.Laplacian(gray, cv2.CV_16S, ksize = 3)Laplacian = cv2.convertScaleAbs(dst)contour = self.find_contours(Laplacian)#提取轮廓点坐标contour_array = contour[0][:, 0, :]#注意这里只保留区域面积最大的轮廓点坐标contours_complex = np.empty(contour_array.shape[:-1], dtype=complex)contours_complex.real = contour_array[:,0]#横坐标作为实数部分contours_complex.imag = contour_array[:,1]#纵坐标作为虚数部分fourier_result = np.fft.fft(contours_complex)#进行傅里叶变换#fourier_result = np.fft.fftshift(fourier_result)descirptor_in_use = self.truncate_descriptor(fourier_result)#截短傅里叶描述子img1 = res.copy()self.reconstruct(res, descirptor_in_use)# 绘图显示描述子点self.draw_circle(img1, descirptor_in_use)# 相关关定位框架return res, descirptor_in_usedef reconstruct(self,img, descirptor_in_use):"""由傅里叶描述子重建轮廓图:param res: 输入图像，傅里叶描述子:return: 重绘图像"""contour_reconstruct = np.fft.ifft(descirptor_in_use)#傅里叶反变换contour_reconstruct = np.array([contour_reconstruct.real,contour_reconstruct.imag])contour_reconstruct = np.transpose(contour_reconstruct)#转换矩阵contour_reconstruct = np.expand_dims(contour_reconstruct, axis = 1)#改变数组维度在axis=1轴上加1if contour_reconstruct.min() < 0:contour_reconstruct -= contour_reconstruct.min()contour_reconstruct *= img.shape[0] / contour_reconstruct.max()contour_reconstruct = contour_reconstruct.astype(np.int32, copy = False)# 中心点M = cv2.moments(contour_reconstruct) # 计算第一条轮廓的各阶矩,字典形式center_x = int(M["m10"] / M["m00"])center_y = int(M["m01"] / M["m00"])black_np = np.ones(img.shape, np.uint8) #创建黑色幕布black = cv2.drawContours(black_np,contour_reconstruct,-1,(255,255,255),3) #绘制白色轮廓black = cv2.circle(black, (center_x, center_y), 4, 255, -1)#绘制中心点cv2.circle(img, (center_x, center_y), 5, 255, -1)#绘制中心点point=[]#二维数组转坐标形式for idx in range(len(contour_reconstruct)):str1=str(contour_reconstruct[idx]).lstrip('[[').rstrip(']]').split(" ")while '' in str1:str1.remove('')point.append((int(str1[0]),int(str1[1])))if point[idx]:cv2.circle(black, point[idx], 3, (0, 255, 255), thickness=-1,lineType=cv2.FILLED)cv2.putText(black, "{}".format(idx),point[idx], cv2.FONT_HERSHEY_SIMPLEX,0.6, (0, 0, 255), 2, lineType=cv2.LINE_AA)#print(contour_reconstruct)print(point)# 凸包hull = cv2.convexHull(contour_reconstruct)# 寻找凸包，得到凸包的角点print("部分凸包信息：")print(hull[0])  # [[194 299]]（坐标）hull2 = cv2.convexHull(contour_reconstruct, returnPoints=False)print(hull2[0])  # [20]（cnt中的索引）print(contour_reconstruct[31])  # [[146  33]]print(cv2.isContourConvex(hull))  # True是否为凸型dist = cv2.pointPolygonTest(contour_reconstruct, (center_x, center_y), True)  # 中心点的最小距离print(dist)cv2.polylines(img, [hull], True, (255,255, 255), 3)# 绘制凸包plt.figure(figsize=(10,10))plt.subplot(1,2,1)plt.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))plt.xlabel(u'凸包轮廓图',fontsize=20)plt.subplot(1,2,2)plt.imshow(cv2.cvtColor(black, cv2.COLOR_BGR2RGB))plt.xlabel(u'傅里叶描述子和重心',fontsize=20)plt.show()#cv2.imshow("contour_reconstruct", img)#cv2.imwrite('recover.png',img)return imgdef draw_circle(self,img, descirptor_in_use):"""获取外接轮廓:param res: 输入图像，傅里叶描述子:return: 重绘图像"""contour_reconstruct = np.fft.ifft(descirptor_in_use)#傅里叶反变换contour_reconstruct = np.array([contour_reconstruct.real,contour_reconstruct.imag])contour_reconstruct = np.transpose(contour_reconstruct)#转换矩阵contour_reconstruct = np.expand_dims(contour_reconstruct, axis = 1)#改变数组维度在axis=1轴上加1if contour_reconstruct.min() < 0:contour_reconstruct -= contour_reconstruct.min()contour_reconstruct *= img.shape[0] / contour_reconstruct.max()contour_reconstruct = contour_reconstruct.astype(np.int32, copy = False)x, y, w, h = cv2.boundingRect(contour_reconstruct)  # 外接矩形cv2.rectangle(img, (x, y), (x + w, y + h), (255, 225,0), 3)rect = cv2.minAreaRect(contour_reconstruct)  # 最小外接矩形box = np.int0(cv2.boxPoints(rect))  # 矩形的四个角点取整cv2.drawContours(img, [box], 0, (0, 255,255), 3)(x, y), radius = cv2.minEnclosingCircle(contour_reconstruct)#最小外接圆(x, y, radius) = np.int0((x, y, radius))  # 圆心和半径取整cv2.circle(img, (x, y), radius, (0,255,0), 2)ellipse = cv2.fitEllipse(contour_reconstruct)#拟合椭圆cv2.ellipse(img, ellipse, (0, 0, 255), 2)df = pd.DataFrame(np.random.rand(10,4), columns = [u'外接矩形',u'最小外接矩阵',u'外接圆',u'椭圆'])fig = df.plot(figsize = (6,6))  #创建图表对象，并复制给figplt.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))plt.xlabel(u'图像轮廓',fontsize=20)    plt.show()return imgdef getHandsPointsByFD(self,rimg,limg):"""获取手势图像:param res: 输入图像，傅里叶描述子:return: 重绘图像"""res1 = self.skinMask(rimg) #进行肤色检测ret1, fourier_right = self.fourierDesciptor(res1)# 傅里叶描述子获取轮廓点res2 = self.skinMask(limg) #ret2, fourier_left = self.fourierDesciptor(res2)#cv2.waitKey(0)cv2.destroyAllWindowsreturn fourier_right,fourier_leftif __name__ == '__main__':print("[INFO]Pose estimation.")img_file = "images/letter/pear.jpg"start = time.time()modelpath = "models/"pose_model = general_pose_model(modelpath)# 1.加载模型print("[INFO]Model loads time: ", time.time() - start)# 骨骼start = time.time()bone_points = pose_model.getBoneKeypoints(img_file) # 2.骨骼关键点print("[INFO]COCO18_Model predicts time: ", time.time() - start)pose_model.vis_bone_pose(img_file, bone_points)# 骨骼连线图、标记图显示list1 = pose_model.getBoneInformation(bone_points)# 3.骨骼特征print("[INFO]Model Bone Information[25]: ", list1)# 手势start = time.time()rimg,limg = pose_model.getHandROI(img_file,bone_points)# 4.拆分左右手图像img1 = rimg.copy()img2 = limg.copy()rhandpoints,lhandpoints = pose_model.getHandsKeypoints(rimg,limg)# 5.手势特征点 By handposeprint("[INFO]Hand_Model predicts time: ", time.time() - start)list2 = pose_model.getHandsInformation(rhandpoints,lhandpoints)print("[INFO]Model Hands Information[20]: ", list2)#fourier_right,fourier_left =pose_model.getHandsPointsByFD(img1,img2)# 特征点 By fourierDesciptor#print("[INFO]fourierDesciptor[32]: ",fourier_right,fourier_left)

Openpose人体骨骼、手势--静态图像标记及分类2（附源码）相关推荐

BERT模型实战之多文本分类（附源码）
BERT模型也出来很久了,之前看了论文学习过它的大致模型(可以参考前些日子写的笔记NLP大杀器BERT模型解读),但是一直有杂七杂八的事拖着没有具体去实现过真实效果如何.今天就趁机来动手写一写实战,顺 ...
【Pytorch】利用Pytorch+GRU实现情感分类（附源码）
在这个实验中,数据的预处理过程以及网络的初始化及模型的训练等过程同前文<利用Pytorch+LSTM实现中文新闻分类>,具体这里就不再重复解释了.如果有读者在对数据集的预处理过程中有疑问, ...
【ML-SVM案例学习】案例一：对鸢尾花数据进行SVM分类（附源码）
文章目录前言一.完整源码分步实现 1.引入库 2.读入数据 3.编码数据 4.数据分割 5.数据SVM分类器构建 6.计算模型的准确率/精度 7.计算决策函数的结构值以及预测值 8.画图总结前 ...
C++日历：不同颜色打印出给定年月的日历，若为本月可高亮标记出当前日期（附源码）
效果图: 功能: 周六周日显示为红色工作日显示为蓝色若查询的是本月,则用绿色标记需要用到的头文件:#include "windows.h" 更改颜色(以绿色为例): SetC ...
Python图像识别实战（四）：搭建卷积神经网络进行图像二分类（附源码和实现效果）
前面我介绍了可视化的一些方法以及机器学习在预测方面的应用,分为分类问题(预测值是离散型)和回归问题(预测值是连续型)(具体见之前的文章). 从本期开始,我将做一个关于图像识别的系列文章,让读者慢慢理解 ...
人体姿态估计(人体关键点检测)2D Pose训练代码和Android源码
人体姿态估计(人体关键点检测)2D Pose训练代码和Android源码目录人体姿态估计(人体关键点检测)2D Pose训练代码和Android源码 1.人体姿态估计2D Pose方法 2.人体姿 ...
基于OpenCV的电影视频人像景别分类算法（源码＆教程）
1.研究背景近年来,随着多媒体技术的高速发展,视频数据也呈现出爆炸性的增长.基于内容的视频检索已成为当前的迫切需求.特别在电影视频领域,单纯的播放已经无法满足用户日益增长的需要.如何准确快速地按照用 ...
基于OpenPose和Human segmentation的游戏人物解析（附源码）
基于OpenPose和Human segmentation的游戏人物解析(附源码) --基于PaddleHub的真人街霸游戏 Github AI studio 街霸(Street Fighter)是大 ...
chainer-人体姿态检测-openpose【附源码】
文章目录前言一.模型构建 1.引入深度学习库库 2.模型构建 3.人体姿态检测 4.效果展示总结前言首先对输入图像进行人体姿态检测本次使用国外已有模型,本人没有做过训练,因为此人脸检 ...

Openpose人体骨骼、手势--静态图像标记及分类2（附源码）

文章目录