pHash-Console/Program.cs

namespace Compare.Test;

using CoenM.ImageHash;
using CoenM.ImageHash.HashAlgorithms;
using Serilog;

public static class Program
{
    public static void Main()
    {
        Log.Logger = new LoggerConfiguration()
            .MinimumLevel.Debug()
            .WriteTo.Console()
            .WriteTo.File(
                "logs/log-.txt",
                rollingInterval: RollingInterval.Day,
                outputTemplate: "{Timestamp:yyyy-MM-dd HH:mm:ss.fff zzz} [{Level:u3}] {Message:lj}{NewLine}{Exception}")
            .CreateLogger();

        try
        {
            Log.Information("Application starting up...");
            RunImageHashing();
            Log.Information("Application has finished successfully.");
        }
        catch (Exception ex)
        {
            Log.Fatal(ex, "An unhandled exception occurred, terminating the application.");
        }
        finally
        {
            Log.CloseAndFlush();
        }
    }

    private static void RunImageHashing()
    {
        // --- Configuration folders ---
        // Takes the folder directories of each contestant
        // Each image should be .PNG and have "XX_" prefix e.g : 01_Image.png, 02_Image2.png, etc
        string[] inputDirectories =
        [
            "/home/ibnufadhil/Documents/projects/StitcherResultTest/Test2.0/renjayarz",
            "/home/ibnufadhil/Documents/projects/StitcherResultTest/Test2.0/reinardras",
            "/home/ibnufadhil/Documents/projects/StitcherResultTest/Test2.0/meizar",
            "/home/ibnufadhil/Documents/projects/StitcherResultTest/Test2.0/olymrifki",
            "/home/ibnufadhil/Documents/projects/StitcherResultTest/Test2.0/fikribahru",
            "/home/ibnufadhil/Documents/projects/StitcherResultTest/Test2.0/benscode",
            "/home/ibnufadhil/Documents/projects/StitcherResultTest/Test2.0/dennisarfan"
        ];

        string outputCsvPath = "hash_analysis_results.csv";
        IImageHash hashAlgorithm = new PerceptualHash();

        Log.Information("Generating hashes using {AlgorithmName}...", hashAlgorithm.GetType().Name);

        var groupedImages = new Dictionary<string, List<string>>();

        // Group images by prefix
        foreach (string dir in inputDirectories)
        {
            if (!Directory.Exists(dir))
            {
                Log.Warning("Directory not found: {DirectoryPath}", dir);
                continue;
            }

            foreach (string file in Directory.GetFiles(dir, "*.png"))
            {
                string prefix = Path.GetFileName(file).Split('_')[0];
                if (!groupedImages.ContainsKey(prefix))
                    groupedImages[prefix] = new List<string>();

                groupedImages[prefix].Add(file);
            }
        }

        var csvLines = new List<string> { "TestPrefix,FileName,ImageHash,SimilarityToRepresentative(%)" };

        foreach (var group in groupedImages.OrderBy(g => g.Key))
        {
            string prefix = group.Key;
            var imagePaths = group.Value;

            if (imagePaths.Count < 2)
            {
                Log.Warning("Skipping {Prefix}: Not enough images to compare ({ImageCount}).", prefix, imagePaths.Count);
                continue;
            }

            Log.Information("=== Processing group {Prefix} ({ImageCount} images) ===", prefix, imagePaths.Count);

            // Calculate hashes
            var hashResults = new Dictionary<string, ulong>();
            foreach (var path in imagePaths)
            {
                try
                {
                    using var stream = File.OpenRead(path);
                    ulong hash = hashAlgorithm.Hash(stream);
                    hashResults[path] = hash;
                    string relativePath = $"{Path.GetFileName(Path.GetDirectoryName(path))}/{Path.GetFileName(path)}";

                    Log.Debug("Hashed {ImagePath}: {ImageHash}", relativePath, hash);
                }
                catch (Exception ex)
                {
                    Log.Error(ex, "Error hashing {ImagePath}", path);
                }
            }

            if (hashResults.Count < 2)
            {
                Log.Warning("Skipping {Prefix}: Not enough valid images after hashing ({HashedCount}).", prefix, hashResults.Count);
                continue;
            }

            // --- Compute Representative Hash ---
            ulong representativeHash = ComputeRepresentativeHash(hashResults.Values.ToList());
            Log.Information("-> Representative Hash for {Prefix}: {RepresentativeHash}", prefix, representativeHash);
            Log.Information("-----------------------------------------------------");

            // --- Compute similarity to representative hash ---
            foreach (var entry in hashResults)
            {
                double similarity = CompareHash.Similarity(entry.Value, representativeHash);
                string fileName = $"{Path.GetFileName(Path.GetDirectoryName(entry.Key))}/{Path.GetFileName(entry.Key)}";
                csvLines.Add($"{prefix},{fileName},{entry.Value},{similarity:F6}");
                Log.Information("{FileName,-70} | Similarity: {Similarity:F2}%", fileName, similarity);
            }
            Log.Information("-----------------------------------------------------\n");
        }

        // --- Write CSV output ---
        try
        {
            File.WriteAllLines(outputCsvPath, csvLines);
            Log.Information("Results successfully saved to: {CsvPath}", Path.GetFullPath(outputCsvPath));
        }
        catch (Exception ex)
        {
            Log.Error(ex, "Failed to write CSV file to {CsvPath}", outputCsvPath);
        }
    }

    // === Compute the bitwise representative hash (majority vote using consensus) ===
    private static ulong ComputeRepresentativeHash(List<ulong> hashes)
    {
        int[] bitCounts = new int[64];

        // Count how many 1s are in each bit
        foreach (ulong hash in hashes)
        {
            for (int bit = 0; bit < 64; bit++)
            {
                if (((hash >> bit) & 1UL) == 1UL)
                    bitCounts[bit]++;
            }
        }

        ulong result = 0UL;
        int threshold = hashes.Count / 2;

        // If more than half of the hashes have a 1 in a bit, set the corresponding bit in the result, otherwise 0
        for (int bit = 0; bit < 64; bit++)
        {
            if (bitCounts[bit] > threshold)
                result |= 1UL << bit;
        }

        return result;
    }
}