旋转图片检测到人脸后计算旋转前的图片的人脸中心点

core/opencv.py

@@ -1,12 +1,13 @@
 """使用Open CV识别图像"""
 import os
-import cv2
+import cv2 as cv
 import sys
+import math
 
 sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
 from core.datatype import mei_path
 
-face_cascade = cv2.CascadeClassifier(mei_path('data/haarcascade_frontalface_default.xml'))
+face_cascade = cv.CascadeClassifier(mei_path('data/haarcascade_frontalface_default.xml'))
 
 
 # https://github.com/PyImageSearch/imutils/blob/master/imutils/convenience.py#L41
@@ -17,7 +18,7 @@ def rotate_bound(image, angle):
 
     # grab the rotation matrix (applying the negative of the angle to rotate clockwise),
     # then grab the sine and cosine (i.e., the rotation components of the matrix)
-    M = cv2.getRotationMatrix2D((cX, cY), -angle, 1.0)
+    M = cv.getRotationMatrix2D((cX, cY), -angle, 1.0)
     cosine = abs(M[0, 0])
     sine = abs(M[0, 1])
 
@@ -30,7 +31,7 @@ def rotate_bound(image, angle):
     M[1, 2] += (nH / 2) - cY
 
     # perform the actual rotation and return the image
-    return cv2.warpAffine(image, M, (nW, nH))
+    return cv.warpAffine(image, M, (nW, nH))
 
 
 def isValidArea(area):
@@ -43,40 +44,68 @@ def isValidArea(area):
         return False
 
 
+def calc_origin_area(img, area, rotate, ori_size):
+    """计算图片中指定区域的中心点在旋转前的位置"""
+    # 计算area的中心点以图片中心点为坐标原点的新坐标
+    (h, w) = img.shape[:2]
+    Ox, Oy = (h//2, w//2)
+    new_x = area[0] + area[2]//2 - Ox
+    new_y = area[1] + area[3]//2 - Oy
+    # 计算area的中心点做反向旋转后的坐标
+    anti_rotate = math.radians(-rotate)
+    anti_x = new_x * math.cos(anti_rotate) - new_y * math.sin(anti_rotate)
+    anti_y = new_x * math.sin(anti_rotate) + new_y * math.cos(anti_rotate)
+    # 计算area的中心点在原图坐标系中的坐标
+    (ori_h, ori_w) = ori_size
+    origin_x = int(ori_w // 2 + anti_x)
+    origin_y = int(ori_h // 2 + anti_y)
+    return (origin_x, origin_y)
+
+
 def detect_faces(img_path, debug=False):
-    img = cv2.imread(img_path)
-    faces = []
+    """检测图片中的人脸并返回第一个有效的人脸区域"""
+    img = cv.imread(img_path)
     # OpenCV的检测效果似乎受图像角度影响很大，因此要尝试旋转图片
     rotate_angles = (0, 10, -10, 20, -20, 30, -30, 40, -40, 50, -50, 60, -60)
     for angle in rotate_angles:
         rotated = rotate_bound(img, angle)
-        grey = cv2.cvtColor(rotated, cv2.COLOR_BGR2GRAY)
+        grey = cv.cvtColor(rotated, cv.COLOR_BGR2GRAY)
         faces = face_cascade.detectMultiScale(grey, 1.1, 4)
         for area in faces:
             if isValidArea(area) and (not debug):
-                return area
+                ori_pos = calc_origin_area(rotated, area, angle, img.shape[:2])
+                return ori_pos
             # 有效区域用绿框标记，无效区域用红框标记
             color = (0, 255, 0) if isValidArea(area) else (0, 0, 255)
             (x, y, w, h) = area
-            cv2.rectangle(rotated, (x, y), (x+w, y+h), color, 2)
+            cv.rectangle(rotated, (x, y), (x+w, y+h), color, 2)
             recog_info = f'({x},{y}), {w}x{h}'
             (h2, w2) = rotated.shape[:2]
             pos = (min(w2-170, x), max(y-10, 12))    # 避免信息显示到绘图区外
-            cv2.putText(rotated, recog_info, pos, cv2.FONT_HERSHEY_SIMPLEX, 0.5, color, 1)
+            cv.putText(rotated, recog_info, pos, cv.FONT_HERSHEY_SIMPLEX, 0.5, color, 1)
         if debug:
             win_name = "Actress detection"
-            cv2.imshow(win_name, rotated)
-            cv2.setWindowTitle(win_name, win_name + f" with ratotion {angle} degrees")
-            cv2.waitKey(0)
-    return faces
+            cv.imshow(win_name, rotated)
+            cv.setWindowTitle(win_name, win_name + f" with ratotion {angle} degrees")
+            cv.waitKey(0)
+
+
+def crop_by_face(img_path, debug=False):
+    """根据指定的人脸区域裁剪图片"""
+    face_center = detect_faces(img_path)
+    if debug:
+        if not os.path.exists('debug'):
+            os.mkdir('debug')
+        img = cv.imread(img_path)
+        cv.circle(img, face_center, 100, (0,255,0), 1)
+        filepath, ext = os.path.splitext(img_path)
+        output = 'debug/' + filepath + '_opencv' + ext
+        cv.imwrite(output, img)
+    return face_center
 
 
 if __name__ == "__main__":
-    import argparse
-    # construct the argument parse and parse the arguments
-    ap = argparse.ArgumentParser()
-    ap.add_argument("-i", "--image", required=True,
-        help="path to the image file")
-    args = vars(ap.parse_args(['-i', 'YPPx.jpg']))
-
-    detect_faces(args['image'], True)
+    if len(sys.argv) < 2:
+        sys.argv.append('siro-4701.jpg')
+    for img_path in sys.argv[1:]:
+        crop_by_face(img_path, True)