From d19abb296c7ddfe4a0b3d0efd027b3194e8ad673 Mon Sep 17 00:00:00 2001
From: zhe <2904040094@qq.com>
Date: Wed, 23 Oct 2024 11:52:36 +0800
Subject: [PATCH] =?UTF-8?q?=E6=9B=B4=E6=96=B0=20function.py?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 function.py | 71 +++++++++++++++++++++++++++++++++--------------------
 1 file changed, 44 insertions(+), 27 deletions(-)

diff --git a/function.py b/function.py
index 103fa23..e63d439 100644
--- a/function.py
+++ b/function.py
@@ -1,39 +1,42 @@
 import numpy as np  
 import os  
 from scipy.spatial.distance import mahalanobis  
-from collections import Counter   
-
-# 学号和姓名  
+from collections import Counter  
+  
+# 学号和姓名（这部分在实际代码中可能不需要，但为了完整性保留）  
 id = "202210423111"  
-name = ""  
-
-def classify(input_vector, dataSet, labels, k=1):  
+name = "李劲"  
+  
+def classify(input_vector, dataSet, labels, mean_vector, inv_cov_matrix, k=1):  
     """  
     KNN 分类算法，使用马氏距离。  
     """  
-    # 计算协方差矩阵及其逆  
-    cov_matrix = np.cov(dataSet.T)  
-    if np.linalg.det(cov_matrix) == 0:  # 检查奇异矩阵  
-        cov_matrix += np.eye(cov_matrix.shape[0]) * 1e-8  # 添加正则化  
-    inv_cov_matrix = np.linalg.inv(cov_matrix)  
-    
-    # 计算马氏距离  
-    distances = [mahalanobis(input_vector, point, inv_cov_matrix) for point in dataSet]  
-    
+    # 确保输入向量与训练数据集的维度一致  
+    input_vector = input_vector.reshape(1, -1)  
+      
+    # 计算每个数据点到输入向量的马氏距离  
+    mahalanobis_distances = []  
+    for data_point in dataSet:  
+        dist = mahalanobis(data_point.reshape(1, -1), input_vector, inv_cov_matrix)  
+        mahalanobis_distances.append(dist)  
+      
+    # 将距离列表转换为 NumPy 数组，方便后续操作  
+    mahalanobis_distances = np.array(mahalanobis_distances)  
+      
     # 获取距离最近的 k 个样本的索引  
-    k_indices = np.argsort(distances)[:k]  
+    k_indices = np.argsort(mahalanobis_distances)[:k]  
       
     # 获取最近邻样本的标签  
     k_nearest_labels = [labels[i] for i in k_indices]  
       
     # 找出出现次数最多的标签  
     most_common = Counter(k_nearest_labels).most_common(1)[0][0]  
-    
+      
     return most_common  
   
 def load_dataset(directory):  
     """  
-    从文件夹中加载数据。  
+    从文件夹中加载数据，并计算均值向量和协方差矩阵。  
     """  
     features = []  
     labels = []  
@@ -51,7 +54,15 @@ def load_dataset(directory):
             except Exception as e:  
                 print(f"Error processing file {filepath}: {e}")  
       
-    return np.array(features), np.array(labels)  
+    features = np.array(features)  
+    labels = np.array(labels)  
+      
+    # 计算均值向量和协方差矩阵  
+    mean_vector = np.mean(features, axis=0)  
+    cov_matrix = np.cov(features, rowvar=False)  
+    inv_cov_matrix = np.linalg.inv(cov_matrix)  # 计算协方差矩阵的逆  
+      
+    return features, labels, mean_vector, inv_cov_matrix  
   
 def img_to_vector(filename):  
     """  
@@ -64,20 +75,26 @@ def img_to_vector(filename):
             return image_vector  
     except Exception as e:  
         print(f"Error reading file {filename}: {e}")  
-        return np.zeros(1024)  # 假设有一个预期的向量长度，此处设为1024  
-
+        return np.zeros(1024)  # 假设图像向量长度为1024  
+  
 if __name__ == "__main__":  
     # 测试  
-    train_data, train_labels = load_dataset("trainingDigits")  
-
-    test_data = (  
+    train_data, train_labels, mean_vector, inv_cov_matrix = load_dataset("trainingDigits")  
+  
+    # 注意：这里的测试数据应该是一个二维数组，形状为 (1, 1024)  
+    # 假设 test_data 是一个已经格式化为字符串的1024位二值图像  
+    test_data_str = (  
         "00000000000000110000000000000000"  
-        # ... (省略的测试数据)  
+        # ... (省略的测试数据，确保总长度为1024)  
     )  
+    test_data = np.array([int(char) for char in test_data_str]).reshape(1, 1024)  
+      
     predicted_label = classify(  
-        np.array([int(char) for char in test_data]).reshape(1, 1024),  
+        test_data,  
         train_data,  
         train_labels,  
+        mean_vector,  
+        inv_cov_matrix,  
     )  
-
+      
     print(f"预测的标签: {predicted_label}")
\ No newline at end of file