import cv2 import numpy as np import os # 提取图像的HOG特征 def get_hog_features(image): hog = cv2.HOGDescriptor() hog_features = hog.compute(image) return hog_features # 加载训练数据集 train_data = [r"I:\18Breakageratecalculation\SVM run\detection_cut\whole\train128"] train_labels = [r"I:\18Breakageratecalculation\SVM run\detection_cut\whole\train128\labels.txt"] num_samples = 681 for i in range(num_samples): img = cv2.imread(str(i).zfill(3)+'.jpg') hog_features = get_hog_features(image) hsv_image = cv2.cvtColor(image, cv2.COLOR_BGR2HSV) color_hist = cv2.calcHist([hsv_image], [0, 1], None, [180, 256], [0, 180, 0, 256]) color_features = cv2.normalize(color_hist, color_hist).flatten() train_data.append(hog_features) train_labels.append(labels[i]) # 训练SVM模型 svm = cv2.ml.SVM_create() svm.setType(cv2.ml.SVM_C_SVC) svm.setKernel(cv2.ml.SVM_LINEAR) svm.train(np.array(train_data), cv2.ml.ROW_SAMPLE, np.array(train_labels)) # 对测试图像进行分类 test_image = cv2.imread('I:\18Breakageratecalculation\mask-slic use\maskSLIC-master\result\split\result2\maskslic2_roi.png', 0) test_features = get_hog_features(test_image) result = svm.predict(test_features.reshape(1,-1)) # 显示分割结果 result_image = np.zeros(test_image.shape, np.uint8) for i in range(test_image.shape[0]): for j in range(test_image.shape[1]): if result[i,j] == 1: result_image[i,j] = 255 cv2.imshow('I:\18Breakageratecalculation\mask-slic use\maskSLIC-master\result\split\result2\Result.png', result_image) cv2.waitKey(0) cv2.destroyAllWindows()

这段代码似乎存在一些问题。以下是一些可能需要修改的地方：

在加载训练数据时，train_data和train_labels应该是列表类型，而不是字符串类型。您可以使用os.listdir()函数来获取文件夹中的所有图像文件名，并使用cv2.imread()函数读取图像文件。

在获取HOG特征时，您应该使用img而不是image变量。

在提取颜色直方图特征时，您应该使用test_image而不是image变量。

在将特征添加到train_data列表中时，您应该将hog_features和color_features链接在一起。

在使用svm.predict()函数对测试图像进行分类时，您应该使用svm.predict(test_features)而不是svm.predict(test_features.reshape(1,-1))。

在显示分割结果时，您应该使用cv2.imshow()函数的第一个参数是窗口的名称，而不是图像文件的路径。

以下是修改后的示例代码：
import cv2
import numpy as np
import os

# 提取图像的HOG特征
def get_hog_features(img):
hog = cv2.HOGDescriptor()
hog_features = hog.compute(img)
return hog_features.flatten()

# 加载训练数据集
train_folder = r"I:\18Breakageratecalculation\SVM run\detection_cut\whole\train128"
train_data = []
train_labels = []
for filename in os.listdir(train_folder):
if filename.endswith('.jpg'):
img = cv2.imread(os.path.join(train_folder, filename), 0)
hog_features = get_hog_features(img)
hsv_image = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
color_hist = cv2.calcHist([hsv_image], [0, 1], None, [180, 256], [0, 180, 0, 256])
color_features = cv2.normalize(color_hist, color_hist).flatten()
features = np.concatenate((hog_features, color_features))
train_data.append(features)
with open(os.path.join(train_folder, 'labels.txt'), 'r') as f:
label = int(f.readline().strip())
train_labels.append(label)

# 训练SVM模型
svm = cv2.ml.SVM_create()
svm.setType(cv2.ml.SVM_C_SVC)
svm.setKernel(cv2.ml.SVM_LINEAR)
svm.train(np.array(train_data), cv2.ml.ROW_SAMPLE, np.array(train_labels))

# 对测试图像进行分类
test_image = cv2.imread('I:\18Breakageratecalculation\mask-slic use\maskSLIC-master\result\split\result2\maskslic2_roi.png', 0)
test_features = get_hog_features(test_image)
hsv_test_image = cv2.cvtColor(test_image, cv2.COLOR_BGR2HSV)
color_hist = cv2.calcHist([hsv_test_image], [0, 1], None, [180, 256], [0, 180, 0, 256])
color_features = cv2.normalize(color_hist, color_hist).flatten()
test_features = np.concatenate((test_features, color_features))
result = svm.predict(test_features)

# 显示分割结果
result_image = np.zeros(test_image.shape, np.uint8)
for i in range(test_image.shape[0]):
for j in range(test_image.shape[1]):
if result[i,j] == 1:
result_image[i,j] = 255
cv2.imshow('Result', result_image)
cv2.waitKey(0)
cv2.destroyAllWindows()
登录后复制
请注意，以上代码仅作为参考。您可能需要根据自己的情况进行微调。