import argparse import os import matplotlib.pyplot as plt import numpy as np import pandas as pd csv_file = "out_all.csv" stability_file = "out_stability.csv" degradation_file = "out_degradation.csv" paper_names = { "1A": "Glossy Pro Platinum (open)", "1B": "Glossy Pro Platinum (close)", "2A": "Photo Fine-Grained Luster (open)", "2B": "Photo Fine-Grained Luster (close)", "3A": "Photo Premium Matte (open)", "3B": "Photo Premium Matte (close)", "4A": "Matte Photo (open)", "4B": "Matte Photo (close)", "5A": "ELECOM Matte Photo (open)", "5B": "ELECOM Matte Photo (close)", } data_dir = ( r"\\gabor\Project\内科\舌診\TIAS\開発\カラーチャート\TCC2025-01\退色調査3 マットフォト" + "\\" ) colors = ["tab:blue", "tab:orange", "tab:green", "tab:red", "tab:purple"] def make_all_data_csv() -> None: """Reads all CSV files.""" # noqa: D401 print("Combining all data into a single CSV file...") files = [f for f in os.listdir(data_dir) if f.endswith(".csv")] all_data = pd.DataFrame() for i, file in enumerate(files): if not file.startswith("tcc6"): continue file_path = os.path.join(data_dir, file) parts = os.path.splitext(os.path.basename(file))[0].split("_") paper = parts[0][5:7] # paper = "4B" # 紙種類固定 datestr = parts[1] + "-" + parts[2] # iter = 0 if file[11:12] == "a" else 1 iter = 0 print(f"Reading {file_path} (Paper: {paper}, Date: {datestr}, Iteration: {iter})") df = pd.read_csv(file_path, encoding="utf-8") df["tcc"] = df["no"].astype("int") df["paper"] = paper df["date"] = datestr df["iter"] = iter # df.drop(columns=["no", "X", "Y", "Z"], inplace=True, errors="ignore") # df.drop(columns=[f"{c:.1f}" for c in range(400, 710, 10)], inplace=True, errors="ignore") all_data = pd.concat([all_data, df], ignore_index=True) if all_data.empty: print("No data found in the specified directory.") exit(1) all_data.sort_values(by=["date", "paper", "iter", "tcc"], inplace=True) # Save the combined DataFrame to a CSV file all_data[["date", "paper", "iter", "tcc", "L*", "a*", "b*"]].to_csv( csv_file, index=False, encoding="utf-8" ) def calc_repeat_stability(): """Calculate repeat stability of color measurements.""" if not os.path.exists(csv_file): make_all_data_csv() print("Calculating repeat stability...") df = pd.read_csv(csv_file, encoding="utf-8", parse_dates=["date"]) no_repeat = False if df["iter"].max() < 1: print("No repeat measurements found.") no_repeat = True df0 = df[df["iter"] == 0].drop(columns=["iter"]).copy() df0.columns = ["date", "paper", "tcc", "L0", "a0", "b0"] if not no_repeat: df1 = df[df["iter"] == 1].drop(columns=["iter"]).copy() df1.columns = ["date", "paper", "tcc", "L1", "a1", "b1"] dfm = pd.merge(df0, df1, on=["date", "paper", "tcc"]) dfm["dL"] = dfm["L1"] - dfm["L0"] dfm["da"] = dfm["a1"] - dfm["a0"] dfm["db"] = dfm["b1"] - dfm["b0"] dfm["dE"] = np.sqrt( dfm["dL"] ** 2 + dfm["da"] ** 2 + dfm["db"] ** 2 ) # Euclidean distance in Lab color space dfm["meanL"] = (dfm["L1"] + dfm["L0"]) / 2 dfm["meana"] = (dfm["a1"] + dfm["a0"]) / 2 dfm["meanb"] = (dfm["b1"] + dfm["b0"]) / 2 else: dfm = df0.copy() dfm["dL"] = 0.0 dfm["da"] = 0.0 dfm["db"] = 0.0 dfm["dE"] = 0.0 dfm["meanL"] = dfm["L0"] dfm["meana"] = dfm["a0"] dfm["meanb"] = dfm["b0"] dfm.to_csv(stability_file, index=False, encoding="utf-8") # print("Top 10 largest deltaE values:") # print(dfm.sort_values("dE", ascending=False).head(10)) # deltaE values Histogram # plt.figure() # dfm["dE"].hist(bins=30) # plt.title("deltaE Histogram") # plt.xlabel("deltaE") # plt.ylabel("Frequency") # plt.show() def color_degradation(): """Analyze color degradation over time.""" if not os.path.exists(stability_file): calc_repeat_stability() print("Calculating color degradation...") dfm = pd.read_csv(stability_file, encoding="utf-8", parse_dates=["date"]) dfm = dfm[["date", "paper", "tcc", "meanL", "meana", "meanb"]] dfm["changeL"] = 0.0 dfm["changea"] = 0.0 dfm["changeb"] = 0.0 dfm["changeDE"] = 0.0 dfm["days"] = 0 paper_list = dfm["paper"].drop_duplicates().sort_values() tcc_list = dfm["tcc"].drop_duplicates().sort_values() dfout = pd.DataFrame() for paper in paper_list: paper_data = dfm[dfm["paper"] == paper] begin_date = paper_data["date"].min() for tcc in tcc_list: tcc_data = paper_data[paper_data["tcc"] == tcc] begin_lab = ( tcc_data[tcc_data["date"] == begin_date][["meanL", "meana", "meanb"]] .iloc[0] .to_list() ) tcc_data.loc[:, "days"] = (tcc_data["date"] - begin_date).dt.total_seconds() // 60 tcc_data.loc[:, "changeL"] = tcc_data["meanL"] - begin_lab[0] tcc_data.loc[:, "changea"] = tcc_data["meana"] - begin_lab[1] tcc_data.loc[:, "changeb"] = tcc_data["meanb"] - begin_lab[2] tcc_data.loc[:, "changeDE"] = np.sqrt( tcc_data["changeL"] ** 2 + tcc_data["changea"] ** 2 + tcc_data["changeb"] ** 2 ) dfout = pd.concat([dfout, tcc_data], ignore_index=True) # print(tcc_data.sort_values("date")) # print("begin_lab", begin_lab) # break # break dfout = dfout.sort_values(by=["paper", "tcc", "date"]).reset_index(drop=True) dfout.to_csv(degradation_file, index=False, encoding="utf-8") def visualize(): """Visualize color degradation over time.""" if not os.path.exists(degradation_file): color_degradation() print("Visualizing color degradation...") df = pd.read_csv(degradation_file, encoding="utf-8", parse_dates=["date"]) paper_list = df["paper"].drop_duplicates().sort_values() fig = plt.figure() fig.set_size_inches(18, 12) fig.suptitle("Color Degradation Over Time (each TCC)", fontsize=16) fig.subplots_adjust(wspace=0.3, hspace=0.4) fig.subplots_adjust(top=0.9, bottom=0.05, left=0.05, right=0.95) for i in range(1, 25): ax = fig.add_subplot(4, 6, i) ax.set_title(f"TCC {i}") ax.set_xlabel("Days") ax.set_ylabel("Change in deltaE") for pidx, paper in enumerate(paper_list): paper_data = df[df["paper"] == paper] if not paper_data.empty: ax.plot( paper_data[paper_data["tcc"] == i]["days"] / (24 * 60), paper_data[paper_data["tcc"] == i]["changeDE"], marker="o" if (pidx % 2) == 0 else "^", markersize=4, color=colors[pidx // 2], label=f"Paper {paper}", ) # ax.set_ylim(0, 3) # Set y-axis limit for better visibility ax.legend() fig.savefig("out_degradation_eachTCC.png", dpi=300, bbox_inches="tight") paper_mean = df.groupby(["paper", "date"]).mean().reset_index() fig2 = plt.figure() fig2.set_size_inches(9, 6) fig2.suptitle("Color Degradation Over Time (mean TCC)", fontsize=16) for pidx, paper in enumerate(paper_list): paper_data = paper_mean[paper_mean["paper"] == paper] if not paper_data.empty: plt.plot( paper_data["days"] / (24 * 60), paper_data["changeDE"], marker="o" if (pidx % 2) == 0 else "^", markersize=6, color=colors[pidx // 2], label=f"Paper {paper} {paper_names[paper]}", ) plt.legend(loc="upper left", borderaxespad=0) plt.xlabel("Days since first measurement") plt.ylabel("Change in deltaE") fig2.savefig("out_degradation_meanTCC.png", dpi=300, bbox_inches="tight") # plt.show() if __name__ == "__main__": # parser = argparse.ArgumentParser(description="Analyze and visualize color degradation.") # parser.add_argument( # "csvfile", # help="Measured CSV file paths", # ) # args = parser.parse_args() # if not args.csvfile.endswith(".csv"): # print("Please provide the path to the CSV file.") # exit(1) # print("datadir:", os.path.dirname(args.csvfile)) # data_dir = os.path.dirname(args.csvfile) print("Starting analysis...") visualize() print("All done.")