diff --git a/classifer.py b/classifer.py
index b45dc6e..f30192c 100644
--- a/classifer.py
+++ b/classifer.py
@@ -461,6 +461,40 @@ class WoodClass(object):
             img = np.clip(img, 0, 255).astype(dtype=np.uint8)
         return img
 
+    def get_luminance_data(self, data_dir=None, plot_2d=False):
+        """
+        获取l数据
+        :param data_dir: 图片路径
+        :param plot_2d: 是否绘制二维图
+        :return:
+        """
+        x_data, y_data, img_names = self.get_train_data(data_dir, plot_2d=plot_2d)
+        x_data = x_data[:, [0, 1, 2]]
+        dark_num = len(x_data[y_data == 0])
+        middle_num = len(x_data[y_data == 1])
+        light_num = len(x_data[y_data == 2])
+        # 将数据按照亮度进行排序
+        x_data = x_data[np.argsort(x_data[:, 0])]
+        # 按照x的顺序，将y和names也进行排序
+        y_data = y_data[np.argsort(x_data[:, 0])]
+        # 按照x的顺序，将img_names也进行排序，但是这里需要注意，img_names是一个list，所以需要先转换成np.array
+        img_names = [img_names[i] for i in np.argsort(x_data[:, 0])]
+
+        # 创建一个labels，用于存储每个像素点的标签
+        labels = np.zeros_like(x_data[:, 0])
+        # 将亮度最低的dark_num个像素点标记为0
+        labels[:dark_num] = 0
+        # 将亮度最高的light_num个像素点标记为2
+        labels[-light_num:] = 2
+        # 将中间的middle_num个像素点标记为1
+        labels[dark_num:-light_num] = 1
+        if plot_2d:
+            plt.figure()
+            plt.scatter(x_data[:, 0], x_data[:, 1], c=labels)
+            plt.show()
+
+        return x_data, y_data, labels, img_names
+
     def get_kmeans_data(self, data_dir=None, plot_2d=False):
         """
         获取kmeans数据
diff --git a/socket_detector.py b/socket_detector.py
index 6a4f8d6..c1f1ada 100644
--- a/socket_detector.py
+++ b/socket_detector.py
@@ -46,7 +46,7 @@ def process_cmd(cmd: str, data: any, connected_sock: socket.socket, detector: Wo
         detector.load(path=settings.model_path)
         response = simple_sock(connected_sock, cmd_type=cmd, result=result)
     elif cmd == 'KM':
-        x_data, y_data, labels, img_names = detector.get_kmeans_data(data, plot_2d=True)
+        x_data, y_data, labels, img_names = detector.get_luminance_data(data, plot_2d=True)
         result = detector.data_adjustments(x_data, y_data, labels, img_names)
         result = ','.join([str(x) for x in result])
         response = simple_sock(connected_sock, cmd_type=cmd, result=result)