画像クラスタリングのコードを書いてる際のエラーを助けていただきたいです。

UnboundLocalError                         Traceback (most recent call last)
Cell In[1], line 122
    118     plot_scatter3d(pca_df)
    121 if __name__ == "__main__":
--> 122     main()

Cell In[1], line 100, in main()
     97 CSV_PATH = r'C:/Users/kuror/image1/microscope_pca.csv'
     99 try:
--> 100     pca_df.to_csv(CSV_PATH, index=False)
    101 except FileNotFoundError:
    102     npy_image_list = load_image(LOAD_PATH)

UnboundLocalError: cannot access local variable 'pca_df' where it is not associated with a value

import glob as gb
import shutil
import cv2
import os
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
from sklearn.preprocessing import StandardScaler
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.ticker as ticker
import seaborn as sns
from mpl_toolkits.mplot3d import Axes3D


# 画像の読み込み
def load_image(path):
    image_list = []
    npy_image_list = []
    image_path_list = gb.glob(f"{path}/*")  # 画像の保存先次第で変更

    for i, image in enumerate(image_path_list):
        image_path = image.replace(chr(92), '/')  # Windows特有の問題を修正
        if i % 100 == 0:
            print(i)
        
        img_npy = cv2.imread(image_path, cv2.IMREAD_COLOR)
        img_npy = cv2.cvtColor(img_npy, cv2.COLOR_BGR2RGB)
        img_npy = cv2.resize(img_npy, (64, 64))
        img_npy = img_npy.flatten()
        npy_image_list.append(img_npy / 255)

    return npy_image_list


def build_kmeans(df, cluster_num):
    kmeans = KMeans(n_clusters=cluster_num, random_state=2021)
    kmeans.fit(df)
    return kmeans


def build_pca(df):
    pca = PCA()
    pca.fit(df)
    return pca


def plot_contribution_rate(pca):
    fig = plt.figure()
    plt.gca().get_xaxis().set_major_locator(ticker.MaxNLocator(integer=True))
    plt.plot([0] + list(np.cumsum(pca.explained_variance_ratio_)), "-o")
    plt.xlabel("Number of principal components")
    plt.ylabel("Cumulative contribution rate")
    plt.grid()
    plt.show()


def plot_scatter2d(df):
    fig = plt.figure()
    sns.scatterplot(data=df, x='PC1', y='PC2', hue='label', palette='bright', legend='full')
    plt.show()


def plot_scatter3d(df):
    fig = plt.figure()
    ax = Axes3D(fig)
    ax.set_xlabel("PC1")
    ax.set_ylabel("PC2")
    ax.set_zlabel("PC3")

    for label in df['label'].values:
        ax.plot(df.loc[df['label'] == label, 'PC1'],
                df.loc[df['label'] == label, 'PC2'],
                df.loc[df['label'] == label, 'PC3'],
                alpha=0.8, marker=".", linestyle='None')
    plt.show()


def make_cluster_dir(load_path, save_path, kmeans):
    shutil.rmtree(save_path, ignore_errors=True)
    os.makedirs(save_path, exist_ok=True)

    for i in range(kmeans.n_clusters):
        cluster_dir = os.path.join(save_path, f"cluster{i}")
        os.makedirs(cluster_dir, exist_ok=True)

    image_path_list = gb.glob(os.path.join(load_path, '*'))
    for label, path in zip(kmeans.labels_, image_path_list):
        shutil.copyfile(path, os.path.join(save_path, f'cluster{label}', os.path.basename(path)))

    print('クラスタごとにファイル作成完了')


def main():
    LOAD_PATH = 'C:/Users/kuror/image1/microscope'
    SAVE_PATH = 'C:/Users/kuror/image1/microscope_clustering'
    CSV_PATH = r'C:/Users/kuror/image1/microscope_pca.csv'
    
    try:
        pca_df.to_csv(CSV_PATH, index=False)
    except FileNotFoundError:
        npy_image_list = load_image(LOAD_PATH)
        df = pd.DataFrame(npy_image_list)
        print(df.shape)
        
        pca = build_pca(df)
        pca_df = pd.DataFrame(pca.transform(df), columns=[f"PC{x + 1}" for x in range(len(df.columns))])
        plot_contribution_rate(pca)
        pca_df.to_csv(CSV_PATH, index=False)
    
    train_df = pca_df.iloc[:, :1200]
    cluster_num = int(input('cluster_num > '))
    kmeans = build_kmeans(train_df, cluster_num)
    make_cluster_dir(LOAD_PATH, SAVE_PATH, kmeans)

    pca_df['label'] = kmeans.labels_
    plot_scatter2d(pca_df)
    plot_scatter3d(pca_df)


if __name__ == "__main__":
    main()