起因
做人脸动作活体检测的时候使用手机拍摄视频, 视频用opencv打开出现了上下颠倒的情况.
当前opencv版本是4.6.0.66, 使用版本4.5.5.64版本正常.
分析
使用ffmpeg查看信息
1 | ffprobe -print_format json -select_streams v -show_streams -i video_20221124_135508.mp4 |
-print_format json 是输出的格式为json. -select_streams v是选择视频流,v是视频, -show_streams 是显示流信息. -i video_20221124_135508.mp4是输入文件地址.
输出结果:1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16"streams": [
{
"index": 0,
"width": 1920,
"height": 1080,
"coded_width": 1920,
"coded_height": 1080,
"side_data_list": [
{
"side_data_type": "Display Matrix",
"rotation": 90
}
]
}
]
}
rotation是一个十进制数, 指定视频在显示之前应逆时针旋转的度数.
这个解释来源于ffmpeg的-display_rotation参数.
还有一个叫rotate也是度数, 官方解释还未找到, 它们的关系在opencv源代码里面可以看出.
rotate等于负的rotation, 如果rotate小于0, 则rotate等于rotate加360度.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20 // https://github.com/opencv/opencv/blob/da4ac6b7eff2e8869567e4faaff73312f9e1ef57/modules/videoio/src/cap_ffmpeg_impl.hpp#L1777
void CvCapture_FFMPEG::get_rotation_angle()
{
rotation_angle = 0;
const uint8_t *data = 0;
data = av_stream_get_side_data(video_st, AV_PKT_DATA_DISPLAYMATRIX, NULL);
if (data)
{
rotation_angle = -cvRound(av_display_rotation_get((const int32_t*)data));
if (rotation_angle < 0)
rotation_angle += 360;
}
AVDictionaryEntry *rotate_tag = av_dict_get(video_st->metadata, "rotate", NULL, 0);
if (rotate_tag != NULL)
rotation_angle = atoi(rotate_tag->value);
}
opencv自带的旋转
搜索opencv的Issues可以知道, 大概是4.5版才加入的默认自动旋转视频, 使用的是ffmpeg的后端.1
2
3
4
5# 默认是自动旋转, 可以设置为0来取消
cap.set(cv2.CAP_PROP_ORIENTATION_AUTO, 0)
# 通过CAP_PROP_ORIENTATION_META可以获取到rotate
orientation = int(cap.get(cv2.CAP_PROP_ORIENTATION_META))
自动旋转的源代码1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24// https://github.com/opencv/opencv/blob/97c6ec6d49cb78321eafe6fa220ff80ebdc5e2f4/modules/videoio/src/cap_ffmpeg.cpp#L144
void rotateFrame(cv::Mat &mat) const
{
bool rotation_auto = 0 != getProperty(CAP_PROP_ORIENTATION_AUTO);
int rotation_angle = static_cast<int>(getProperty(CAP_PROP_ORIENTATION_META));
if(!rotation_auto || rotation_angle%360 == 0)
{
return;
}
cv::RotateFlags flag;
if(rotation_angle == 90 || rotation_angle == -270) { // Rotate clockwise 90 degrees
flag = cv::ROTATE_90_CLOCKWISE;
} else if(rotation_angle == 270 || rotation_angle == -90) { // Rotate clockwise 270 degrees
flag = cv::ROTATE_90_COUNTERCLOCKWISE;
} else if(rotation_angle == 180 || rotation_angle == -180) { // Rotate clockwise 180 degrees
flag = cv::ROTATE_180;
} else { // Unsupported rotation
return;
}
cv::rotate(mat, mat, flag);
}
自定义旋转
如果是其他的opencv不支持旋转的后端, 可以自己实现.1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27# 方式1
cv2.rotate()
# 方式2
cv2.flip()
# 方式3
def rotate_bound(image, angle):
# grab the dimensions of the image and then determine the
# center
(h, w) = image.shape[:2]
(cX, cY) = (w // 2, h // 2)
# 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)
# getRotationMatrix2D, angle正值表示逆时针旋转（坐标原点假定为左上角）
M = cv2.getRotationMatrix2D((cX, cY), -angle, 1.0)
cos = np.abs(M[0, 0])
sin = np.abs(M[0, 1])
# compute the new bounding dimensions of the image
nW = int((h * sin) + (w * cos))
nH = int((h * cos) + (w * sin))
# adjust the rotation matrix to take into account translation
M[0, 2] += (nW / 2) - cX
M[1, 2] += (nH / 2) - cY
# perform the actual rotation and return the image
return cv2.warpAffine(image, M, (nW, nH))