树莓派电容式触摸屏试用——瞳孔跟踪
本帖最后由 云天 于 2022-5-2 16:39 编辑【项目背景】
4月份获得了试用树莓派5寸TFT电容式触摸屏的机会，到手后，因工作原因，一直没有去过。这两天有点时间，看了一下Mediapipe官网，发现了“MediaPipe Iris”功能，
可惜，MediaPipe SDK的 python版本是不支持虹膜检测。
这个难不住我。
【项目设计】
先通过MediaPipe的人脸landmark提供了468个点位的人脸点云数据，
根据编号，很容筛选出左眼与右眼所在的区域所有点，然后完成区域的截取。
再使用Opencv的cv2.findContours实现瞳孔识别与跟踪。
【程序代码】
1、加载Mediapipe和OpenCV库
import cv2 as cv2
importmediapipe as mp
import numpy as np2、人脸landmark检测代码如下：
# 导入可视化函数和可视化样式
mp_drawing=mp.solutions.drawing_utils
#mp_drawing_styles=mp.solutions.drawing_styles
# 导入三维人脸关键点检测模型
mp_face_mesh=mp.solutions.face_mesh
drawing_spec = mp_drawing.DrawingSpec(thickness=1, circle_radius=1)
cap = cv2.VideoCapture(0)
with mp_face_mesh.FaceMesh(
max_num_faces=4,
min_detection_confidence=0.5,
min_tracking_confidence=0.5) as face_mesh:
while cap.isOpened():
success, frame = cap.read()
if not success:
print("Ignoring empty camera frame.")
# If loading a video, use 'break' instead of 'continue'.
break
h, w, c = frame.shape
# image = cv2.resize(frame, (w //2, h//2))
# frame = cv2.flip(frame, 1)
image = np.copy(frame)
h2, w2, c2 = image.shape
# To improve performance, optionally mark the image as not writeable to
# pass by reference.
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
results = face_mesh.process(image)
# Draw the face mesh annotations on the image.
image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
left_eyes = []
right_eyes = []
if results.multi_face_landmarks:
for face_landmarks in results.multi_face_landmarks:
for idx, landmark in enumerate(face_landmarks.landmark):
if idx == 246 or idx == 159 or idx == 158or idx == 145 or idx == 153 or idx == 190:# left
#if idx == 53 or idx == 111:# left
x1 = np.int(landmark.x * w2)
y1 = np.int(landmark.y * h2)
left_eyes.append((x1, y1))
#cv2.circle(image, (x1, y1), 4, (255, 0, 255), 4, cv2.LINE_AA)
if idx ==374 or idx == 380or idx == 385 or idx == 386 or idx == 390or idx == 414: # right
#if idx == 276 or idx == 340: # right
x1 = np.int(landmark.x * w2)
y1 = np.int(landmark.y * h2)
right_eyes.append((x1, y1))
#cv2.circle(image, (x1, y1), 4, (0, 255, 255), 4, cv2.LINE_AA)
if len(right_eyes)+len(left_eyes)==12:
#矩形边框（Bounding Rectangle）是说，用一个最小的矩形，把找到的形状包起来。
right_box = cv2.boundingRect(np.asarray(right_eyes))
left_box = cv2.boundingRect(np.asarray(left_eyes))
detect_iris(image, right_box, left_box)
cv2.imshow('MediaPipe Face Mesh', image)
if cv2.waitKey(5) & 0xFF == 27:
cv2.imwrite("D:/iris_detect_result.png", image)
break
cap.release()
cv2.waitKey(0)
cv2.destroyAllWindows()
3、虹膜l检测代码如下：
def detect_iris(image, right_box, left_box):
left_roi = image:left_box + left_box, left_box:left_box + left_box]
cv2.imshow('left_eye', left_roi)
lh, lw, lc = left_roi.shape
right_roi = image:right_box+right_box,right_box:right_box+right_box]
cv2.imshow('right_eye', right_roi)
rh, rw, rc = right_roi.shape
if rh>0 and lh>0:
rows, cols, _ =right_roi.shape #保存视频尺寸以备用
gray_roi = cv2.cvtColor(right_roi, cv2.COLOR_BGR2GRAY) #转灰度
gray_roi = cv2.GaussianBlur(gray_roi, (7, 7), 0) #高斯滤波一次
_, threshold = cv2.threshold(gray_roi, 30, 255, cv2.THRESH_BINARY_INV)#二值化，依据需要改变阈值
contours, _ = cv2.findContours(threshold, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE) #画连通域
contours = sorted(contours, key=lambda x: cv2.contourArea(x), reverse=True)
for cnt in contours:
(x, y, w, h) = cv2.boundingRect(cnt)
cv2.circle(right_roi, (x + int(w/2), y + int(h/2)), int(h/2), (0, 0, 255), 3)
break
rows, cols, _ =left_roi.shape #保存视频尺寸以备用
gray_roi = cv2.cvtColor(left_roi, cv2.COLOR_BGR2GRAY) #转灰度
gray_roi = cv2.GaussianBlur(gray_roi, (7, 7), 0) #高斯滤波一次
_, threshold = cv2.threshold(gray_roi, 30, 255, cv2.THRESH_BINARY_INV)#二值化，依据需要改变阈值
contours, _ = cv2.findContours(threshold, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE) #画连通域
contours = sorted(contours, key=lambda x: cv2.contourArea(x), reverse=True)
for cnt in contours:
(x, y, w, h) = cv2.boundingRect(cnt)
cv2.circle(left_roi, (x + int(w/2), y + int(h/2)), int(h/2), (0, 0, 255), 3)
break
【硬件连接】
【测试图片】
【演示视频】
https://www.bilibili.com/video/BV1xZ4y1y7jQ?share_source=copy_web
貌似懂了电影里的特效之一~ 占个板凳 666666666666666 666，很科幻 6666666666666666666 大写的佩服！！！！！ 666666666666 {:7_218:}，，
页:
[1]