如何用OpenCV为含细节的蓝色区域及黄色容器图像绘制准确边界框？

1. Blue Container: Fixing Internal Detail Interference #

The core issue is internal details breaking contour detection. Instead of blurring (which erodes edge clarity), use color masking to isolate the blue region directly—this eliminates internal distractions while preserving sharp edges.

Implementation Code #

import cv2
import numpy as np

img = cv2.imread('blue_object.jpg')
hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)

# Adjust these HSV values to match your exact blue container
lower_blue = np.array([90, 50, 50])
upper_blue = np.array([130, 255, 255])

# Create mask to extract only the blue area
mask = cv2.inRange(hsv, lower_blue, upper_blue)
# Clean up mask: remove small noise and fill gaps
mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, np.ones((5,5), np.uint8))
mask = cv2.morphologyEx(mask, cv2.MORPH_OPEN, np.ones((5,5), np.uint8))

# Detect only outer contours to ignore internal details
contours, hierarchy = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

for cnt in contours:
    area = cv2.contourArea(cnt)
    if area > 50000:  # Keep only large contours (adjust threshold as needed)
        x,y,w,h = cv2.boundingRect(cnt)
        cv2.rectangle(img,(x,y),(x+w,y+h),(0,255,0),2)

cv2.imshow('img', img)
cv2.waitKey(0)
cv2.destroyAllWindows()

Key Notes #

• Use a color picker tool to fine-tune the HSV range for your specific blue shade.
• RETR_EXTERNAL ensures we only detect the outer contour of the container, ignoring internal edges from hinges or text.

2. Yellow Container: Separating from Tree Background #

Your current HSV range is too broad, including tree colors. Narrow the range and add contour filtering to isolate the container.

Implementation Code #

import cv2
import numpy as np

img = cv2.imread('yellow_container.jpg')
hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)

# Refined HSV range for yellow (adjust based on your image)
lower_yellow = np.array([20, 100, 100])
upper_yellow = np.array([30, 255, 255])

mask = cv2.inRange(hsv, lower_yellow, upper_yellow)
# Remove small tree fragments from the mask
mask = cv2.morphologyEx(mask, cv2.MORPH_OPEN, np.ones((7,7), np.uint8))
# Fill gaps in the container mask
mask = cv2.dilate(mask, np.ones((5,5), np.uint8), iterations=2)

contours, hierarchy = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

# Filter out small contours (tree parts) by area
for cnt in contours:
    area = cv2.contourArea(cnt)
    if area > 80000:  # Increase threshold to exclude tiny tree regions
        x,y,w,h = cv2.boundingRect(cnt)
        cv2.rectangle(img,(x,y),(x+w,y+h),(0,255,0),2)

cv2.imshow('img', img)
cv2.waitKey(0)
cv2.destroyAllWindows()

Key Notes #

• Use OpenCV trackbars to adjust HSV values in real-time for precise masking.
• MORPH_OPEN erodes then dilates the mask, eliminating small white noise from tree pixels.
• A higher area threshold ensures only the large container contour is selected.

General Tips #

• For edge clarity, apply a light Gaussian blur (cv2.GaussianBlur(img, (3,3), 0)) before masking to reduce high-frequency noise without losing edge definition.
• Always test contour retrieval modes (RETR_EXTERNAL vs RETR_TREE) to find the best fit for your use case.

内容的提问来源于stack exchange，提问作者steve