900字范文 > python+ opencv实现摄像头实时人脸识别并实现汉字标框

python+ opencv实现摄像头实时人脸识别并实现汉字标框

时间：2024-08-16 05:56:36

opencv的puttxt（）函数不能汉字输出，这也是困惑好多人都问题，经过几天的查资料，改代码终于成功实现opencv汉字输出。

第一种方法是是通过写一段代码，能够转码，封装一下再调用，从而实现汉字输出。

第二种方法是使用PIL进行转换一下

以下这个是ft2.py 实现转码的代码

# -*- coding: utf-8 -*-# /zizi7/article/details/70145150import numpy as npimport freetypeimport copyimport pdbclass put_chinese_text(object):def __init__(self, ttf):self._face = freetype.Face(ttf)def draw_text(self, image, pos, text, text_size, text_color):'''draw chinese(or not) text with ttf:param image:image(numpy.ndarray) to draw text:param pos: where to draw text:param text:the context, for chinese should be unicode type:param text_size: text size:param text_color:text color:return:image'''self._face.set_char_size(text_size * 64)metrics = self._face.sizeascender = metrics.ascender/64.0#descender = metrics.descender/64.0#height = metrics.height/64.0#linegap = height - ascender + descenderypos = int(ascender)if not isinstance(text, str):#对于Python 2中的unicode和Python 3中的str，对于Python 2中的str/bytes和Python 3中的bytes的二进制文件text = text.decode('utf-8')img = self.draw_string(image, pos[0], pos[1]+ypos, text, text_color)return imgdef draw_string(self, img, x_pos, y_pos, text, color):'''draw string:param x_pos: text x-postion on img:param y_pos: text y-postion on img:param text: text (unicode):param color: text color:return:image'''prev_char = 0pen = freetype.Vector()pen.x = x_pos << 6 # div 64pen.y = y_pos << 6hscale = 1.0matrix = freetype.Matrix(int((hscale)*0x10000), int(0.2*0x10000),\int(0.0*0x10000), int(1.1*0x10000))cur_pen = freetype.Vector()pen_translate = freetype.Vector()image = copy.deepcopy(img)for cur_char in text:self._face.set_transform(matrix, pen_translate)self._face.load_char(cur_char)kerning = self._face.get_kerning(prev_char, cur_char)pen.x += kerning.xslot = self._face.glyphbitmap = slot.bitmapcur_pen.x = pen.xcur_pen.y = pen.y - slot.bitmap_top * 64self.draw_ft_bitmap(image, bitmap, cur_pen, color)pen.x += slot.advance.xprev_char = cur_charreturn imagedef draw_ft_bitmap(self, img, bitmap, pen, color):'''draw each char:param bitmap: bitmap:param pen: pen:param color: pen color e.g.(0,0,255) - red:return: image'''x_pos = pen.x >> 6y_pos = pen.y >> 6cols = bitmap.widthrows = bitmap.rowsglyph_pixels = bitmap.bufferfor row in range(rows):for col in range(cols):if glyph_pixels[row*cols + col] != 0:img[y_pos + row][x_pos + col][0] = color[0]img[y_pos + row][x_pos + col][1] = color[1]img[y_pos + row][x_pos + col][2] = color[2]if __name__ == '__main__':# just for testimport cv2line = '你好'img = np.zeros([300,300,3])color_ = (0,255,0) # Greenpos = (3, 3)text_size = 24#ft = put_chinese_text('wqy-zenhei.ttc')ft = put_chinese_text('msyh.ttf')image = ft.draw_text(img, pos, line, text_size, color_)cv2.imshow('diplay', image)cv2.waitKey(0)

里面包含了几个函数，其中需要一个字体msyh.ttf在同一文件夹下

（有时候这种方法实现不了，也可以使用PIL库转换的方法，不过那种方法的缺点在于需要加载库，不过简单易懂）

以下是实现人脸识别的代码

# -*- coding: utf-8 -*-# 摄像头头像识别import face_recognitionimport cv2import ft2source = "rtsp://admin:5417010101xx@192.168.1.61/Streaming/Channels/1"cam = cv2.VideoCapture(source)# 本地图像zwh_image = face_recognition.load_image_file("zwh.jpg")zwh_face_encoding = face_recognition.face_encodings(zwh_image)[0]# 本地图像二chenduling_image = face_recognition.load_image_file("chenduling.jpg")chenduling_face_encoding = face_recognition.face_encodings(chenduling_image)[0]# 本地图片三liujunbo_image = face_recognition.load_image_file("liujunbo.jpg")liujunbo_face_encoding = face_recognition.face_encodings(liujunbo_image)[0]# Create arrays of known face encodings and their names# 脸部特征数据的集合known_face_encodings = [zwh_face_encoding,chenduling_face_encoding,liujunbo_face_encoding]# 人物名称的集合known_face_names = ["张文豪","陈都灵","刘军波"]face_locations = []face_encodings = []face_names = []process_this_frame = Truewhile(cam.isOpened()):# 读取摄像头画面ret, frame = cam.read()if not ret:#等同于 if ret is not nonebreak# 改变摄像头图像的大小，图像小，所做的计算就少small_frame = cv2.resize(frame, (0, 0), fx=0.33, fy=0.33)# opencv的图像是BGR格式的，而我们需要是的RGB格式的，因此需要进行一个转换。rgb_small_frame = small_frame[:, :, ::-1]# Only process every other frame of video to save timeif process_this_frame:# 根据encoding来判断是不是同一个人，是就输出true，不是为flaseface_locations = face_recognition.face_locations(rgb_small_frame)face_encodings = face_recognition.face_encodings(rgb_small_frame, face_locations)face_names = []for face_encoding in face_encodings:# 默认为unknownmatches = pare_faces(known_face_encodings, face_encoding,tolerance=0.48)#阈值太低容易造成无法成功识别人脸，太高容易造成人脸识别混淆默认阈值tolerance为0.6#print(matches)name = "Unknown"# if match[0]:#name = "michong"# If a match was found in known_face_encodings, just use the first one.if True in matches:first_match_index = matches.index(True)name = known_face_names[first_match_index]face_names.append(name)process_this_frame = not process_this_frame# 将捕捉到的人脸显示出来for (top, right, bottom, left), name in zip(face_locations, face_names):# Scale back up face locations since the frame we detected in was scaled to 1/4 size#由于我们检测到的帧被缩放到1/4大小，所以要缩小面位置top *= 3right *= 3bottom *= 3left *= 3# 矩形框cv2.rectangle(frame, (left, top), (right, bottom), (0, 0, 255), 3)ft = ft2.put_chinese_text('msyh.ttf')#引入ft2中的字体#加上标签#cv2.rectangle(frame, (left, bottom - 20), (right, bottom), (0, 0, 255), cv2.FILLED)# cv2.putText(frame, name, (left + 6, bottom - 6), font, 0.8, (255, 255, 255), 1)这是不输入汉字时可以用的代码frame = ft.draw_text(frame, (left+10 , bottom ), name, 20, (255, 255, 255))#cv2.imshow("frame",image)会出来两个框一个monitor 一个frame后者显示image但只有frame为true时才会显示#def draw_text(self, image, pos, text, text_size, text_color)cv2.imshow('monitor', frame)if cv2.waitKey(1) & 0xFF == 27:breakcam.release()cv2.destroyAllWindows()

使用PIL转换由于cv2和PIL中颜色的hex码储存顺序不同，需要进行转换下面这段代码可以替换上面的汉字标框的方法。

cv2img = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB) # cv2和PIL中颜色的hex码的储存顺序不同pilimg = Image.fromarray(cv2img)draw = ImageDraw.Draw(pilimg) # 图片上打印font = ImageFont.truetype("msyh.ttf", 27, encoding="utf-8") # 参数1：字体文件路径，参数2：字体大小draw.text((left+10 , bottom ), name, (220, 20, 60), font=font) # 参数1：打印坐标，参数2：文本，参数3：字体颜色，参数4：字体# PIL图片转cv2 图片frame = cv2.cvtColor(np.array(pilimg), cv2.COLOR_RGB2BGR)

最基本的要装上python和opencv，距离远不能看出人脸的解决方法：可以通过调整获取每帧图像的大小可以进行改造，代价就是会变卡，延迟高，这就需要通过调整摄像头的参数进行修改分辨率视频帧率码率上限视频编码来进行降低分辨率。

人脸混淆较近看不出人脸的解决方法：修改阈值阈值太低容易造成无法成功识别人脸，太高容易造成人脸识别混淆默认阈值tolerance为0.6 亚洲人一般要用低点