我可以使用哪些技术来改进微弱阴影的检测？ [已关闭]

Question

我每 10 分钟拍摄一次放在太阳前面的 10 厘米钉子阴影的图像。然而，我在清晨和傍晚时面临着挑战，当时阴影变得微弱而苍白，使我的算法很难检测到它们。这是我试图检测的微弱阴影的示例：

当阴影颜色丰富或强烈时，我的算法成功检测到它。然而，在阴影较弱的情况下，它无法识别该线。我需要获取线的顶部和底部的坐标来计算其长度。

我的代码是：

import time
import numpy as np
import cv2
from math import atan, sqrt, degrees

def captureImage():
    print('Capturing image')
    videoCaptureObject = cv2.VideoCapture(1)

    result = True
    while(result):
        ret, frame = videoCaptureObject.read()
        cv2.imwrite("Newpicture.jpg", frame)
        result = False
    videoCaptureObject.release()
    return frame

def processImage(im):
    print('Processing image')
    image = im

    # Convert image to grayscale
    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    
    # Apply Gaussian blur to reduce noise
    blurred = cv2.GaussianBlur(gray, (5, 5), 0)
    
    # Apply a median filter to enhance the line
    blurred = cv2.medianBlur(blurred, 5)
    
    # Convert grayscale image to binary using a manually adjusted threshold
    _, thresh = cv2.threshold(blurred, 100, 255, cv2.THRESH_BINARY_INV)
    
    # Apply edge detection
    edges = cv2.Canny(thresh, 50, 150)
    
    # Find contours in the edge-detected image
    contours, _ = cv2.findContours(edges, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    
    if contours:
        # Get the largest contour (assuming it's the line)
        c = max(contours, key=cv2.contourArea)
        
        # Get the extreme points of the contour (line)
        x1, y1 = c[c[:, :, 0].argmin()][0]
        x2, y2 = c[c[:, :, 0].argmax()][0]
        
        # Calculate the length of the line
        length = sqrt((x2 - x1)**2 + (y2 - y1)**2)
        
        # Print the coordinates and length
        print(f"Top: ({x1}, {y1}), Bottom: ({x2}, {y2}), Length: {length}")

# Repeat the process 100 times with a 3-minute interval
for i in range(100):
    captureImage()
    time.sleep(180)  # 3 minutes in seconds
    processImage(captureImage())

我调整了高斯模糊和边缘检测的参数，但这些更改并没有解决问题。我什至购买了更高质量的相机，但问题仍然存在。

Answer 1

您可以尝试使用自适应阈值来处理照明的变化，沿着这些思路：

import cv2 as cv

# Load image as greyscale
img = cv.imread('line.jpg', cv.IMREAD_GRAYSCALE)

# Threshold relative to brightness of local 49x49 area
th = cv.adaptiveThreshold(img,255,cv.ADAPTIVE_THRESH_MEAN_C, cv.THRESH_BINARY_INV,49,10)

# Save result
cv.imwrite('result.png', th)