Adds mutators and experiment setup

Calculates basic phashes for images and their mutations.
More stats are needed, but it's a start

src/bin/dct.rs

@@ -1,66 +0,0 @@
-use std::env;
-use std::f64::consts::{PI, SQRT_2};
-use image::{GenericImageView, GrayImage};
-
-fn dct(img: &image::DynamicImage) -> GrayImage {
-	//let mut dct_values: = [0.0; img.width() * img.height()];
-	let w = img.width();
-	let h = img.height();
-	let mut dct_img = GrayImage::new(w, h);
-
-	//let mut dct_values: Vec<f64> = Vec::new();
-	for u in 0..w {
-		for v in 0..h {
-			let w = w as f64;
-			let h = h as f64;
-			//let k = (v*img.width())+u;
-			let mut alpha = 0.25;
-			if u == 0 {
-				alpha = alpha / SQRT_2
-			}
-			if v == 0 {
-				alpha = alpha / SQRT_2
-			}
-			let v: f64 = v as f64;
-			let u: f64 = u as f64;
-			let mut sum: f64 = 0.0;
-			for (x, y, pix) in img.pixels() {
-				let x: f64 = x as f64;
-				let y: f64 = y as f64;
-				let pixel = pix[0] as f64;
-				// sum +=
-				// 	pixel * 
-				// 	(x*u*PI/16.0).cos() *
-				// 	(y*v*PI/16.0).cos()
-				sum += 
-					pixel *
-					(
-						PI/w*
-						(x+0.5)*u
-					).cos()
-					*
-					(
-						PI/h*
-						(y+0.5)*v
-					).cos()
-			}
-			//dct_values[k] = alpha * sum;
-			//dct_values.push(alpha * sum);
-			print!("{sum}\t");
-			//sum = (sum + 200000.0)/784.0;
-			print!("{}\n", sum * alpha);
-			let pixel = (sum).abs() as u8;
-			println!("{u}, {v}:\t {sum}");
-			dct_img.put_pixel(u as u32, v as u32, image::Luma([pixel]));
-		}
-	}
-	dct_img
-}
-
-fn main() {
-    let args: Vec<String> = env::args().collect();
-	let path = args[1].clone();
-	let img = image::open(&path).unwrap();
-	let dct = dct(&img);
-	dct.save("dct.png").unwrap();
-}

src/bin/mutate.rs

@@ -0,0 +1,38 @@
+use image_similarity::mutators::get_all_mutators;
+use clap::Parser;
+
+#[derive(Parser)]
+#[command(version, about, long_about = None)]
+struct Cfg {
+	/// Path to the image
+	path: String,
+	/// Try to output the images to terminal with viuer
+	#[arg(short, long, default_value_t=false)]
+	show_images: bool,
+}
+
+fn main() {
+	env_logger::init();
+
+	let cfg = Cfg::parse();
+
+	let img = image::open(&cfg.path)
+		.expect("Unable to open file");
+
+	let mutators = get_all_mutators();
+
+	let conf = viuer::Config {
+	    width: Some(16),
+	    height: Some(8),
+	    ..Default::default()
+	};
+	for mutator in mutators {
+		let mutated = mutator.mutate(&img);
+		let filename = "mut".to_string() + &mutator.tag() + &"jpg".to_string();
+		println!("Saving to {}", filename);
+		mutated.save(filename).expect("Saving image failed");
+		if cfg.show_images {
+			viuer::print(&mutated, &conf).expect("Image printing failed.");
+		}
+	}
+}

src/descriptors/dct.rs

@@ -104,7 +104,7 @@ impl DCT {
 
 impl Descriptor for DCT {
     fn info(&self) -> String {
-        "DCT".to_string()
+        "dct".to_string()
     }
     fn describe(&self, img: &image::DynamicImage) -> u64 {
         let resized = self.resize(img);

src/descriptors/mod.rs

@@ -1,6 +1,8 @@
 //! Descriptors that can be used to describe an image.
 //! If two images are (almost) the same, their descriptions will be the same.
 
+use image::GenericImageView;
+
 pub trait Descriptor {
     /// Print the name of the descriptor, 
     fn info(&self) -> String;
@@ -36,34 +38,44 @@ pub struct DCT {
 }
 pub mod dct;
 
-// fn median64<T: Ord + Copy>(values: &[T]) -> T {
+fn median64<T: Ord + Copy>(values: &[T]) -> T {
-//     let mut sorted_values = values.to_vec();
+    let mut sorted_values = values.to_vec();
-//     sorted_values.sort();
+    sorted_values.sort();
-//     let len = sorted_values.len();
+    let len = sorted_values.len();
-//     sorted_values[len/2]
+    sorted_values[len/2]
-// }
+}
 
+pub struct Median;
 
-// pub struct Median;
+impl Descriptor for Median {
+    fn info(&self) -> String {
+        "median".to_string()
+    }
 
-// impl Descriptor for Median {
+    fn describe(&self, img: &image::DynamicImage) -> u64 {
-//     fn describe(&self, img: image::DynamicImage) -> u64 {
+        let img = self.resize(img);
-//         let img = self.resize(img);
+        let mut values: [u8; 64] = [0; 64];
-//         let mut values: [u8; 64] = [0; 64];
+        let mut i: usize = 0;
-//         let mut i: usize = 0;
+        for (_, _, pix) in img.pixels() {
-//         for (_, _, pix) in img.pixels() {
+            values[i] = pix[0];
-//             values[i] = pix[0];
+            i = i+1;
-//             i = i+1;
+        }
-//         }
+        let median = median64(&values);
-//         let median = median64(&values);
+        let mut mask: u64 = 0;
-//         let mut mask: u64 = 0;
+        img.save("debug.png").unwrap();
-//         img.save("debug.png").unwrap();
+        for (_, _, pix) in img.pixels() {
-//         for (_, _, pix) in img.pixels() {
+            if pix[0] > median {
-//             if pix[0] > median {
+                mask += 1;
-//                 mask += 1;
+            }
-//             }
+            mask = mask << 1;
-//             mask = mask << 1;
+        }
-//         }
+        mask
-//         mask
+    }
-//     }
+}
-// }
+
+pub fn get_all_descriptors() -> Vec<Box<dyn Descriptor>> {
+	let mut descriptors: Vec<Box<dyn Descriptor>> = Vec::with_capacity(2);
+	descriptors.push(Box::new(DCT::new()));
+	descriptors.push(Box::new(Median));
+	descriptors
+}

src/lib.rs

@@ -4,3 +4,4 @@
 //! If two images are (almost) the same, their descriptions will be the same.
 pub mod descriptors;
 pub mod store;
+pub mod mutators;

src/main.rs

@@ -1,101 +1,63 @@
 use image_similarity::store::DescriptorStore;
-use image_similarity::descriptors::{Descriptor, DCT};
+use image_similarity::descriptors::get_all_descriptors;
-use log::info;
+use image_similarity::mutators::get_all_mutators;
+use log::{error, info};
 use std::path::PathBuf;
-use std::thread;
+use std::{fs, thread};
 use clap::Parser;
 
 #[derive(Parser)]
 #[command(version, about, long_about = None)]
 struct Cfg {
     /// Path that contains the input images. Will not traverse directories.
-    path: PathBuf
+    path: PathBuf,
-}
-
-fn make_store<T: Descriptor>(descriptor: T, input_path: PathBuf, save_path: PathBuf) -> DescriptorStore<T> {
-    let mut store =
-        DescriptorStore::new(descriptor)
-        .with_file(save_path);
-    info!("{}", store);
-    store.insert_directory(input_path);
-    info!("{}", store);
-    store.save().expect("Error saving");
-    store
-}
-
-struct StoreParams {
-    desc: DCT,
-    input_path: PathBuf,
-    save_path: PathBuf,
 }
 
 fn main() {
     env_logger::init();
     let cfg = Cfg::parse();
+	let mutators = get_all_mutators();
+	//let descriptors = get_all_descriptors();
+    let mut stores: Vec<DescriptorStore> = Vec::new();
 
-    let mut threads = vec![];
+    for descriptor in get_all_descriptors() {
-
+        let store = DescriptorStore::new(descriptor);
-    let params = vec![
+        stores.push(store);
-        StoreParams {
-            desc: DCT::new().with_quality(50),
-            input_path: cfg.path.clone(),
-            save_path: "dct50.messagepack".into(),
-        },
-        // StoreParams {
-        //     desc: DCT::new().with_quality(30),
-        //     input_path: cfg.path.clone(),
-        //     save_path: "dct30.messagepack".into(),
-        // },
-        // StoreParams {
-        //     desc: DCT::new().with_quality(10),
-        //     input_path: cfg.path.clone(),
-        //     save_path: "dct10.messagepack".into(),
-        // },
-    ];
-
-    // let dct_50 =
-    //     DCT::new()
-    //     .with_quality(50);
-    // let dct_30 =
-    //     DCT::new()
-    //     .with_quality(30);
-    // let dct_10 =
-    //     DCT::new()
-    //     .with_quality(10);
-
-    //let (tx, rx) = mpsc::channel::<DescriptorStore>();
-
-    for param in params {
-        threads.push(
-            thread::spawn(move || {
-                let store = make_store(param.desc, param.input_path, param.save_path);
-                store.print_most_dups();
-                //tx.send(store).unwrap();
-            })
-        )
     }
     
-    // let path_50 = cfg.path.clone();
+    for node in fs::read_dir(cfg.path).unwrap() {
-    // threads.push(
+        println!("{:?}", node);
-    //     thread::spawn(move || {
+        let file = node.expect("Error walking directory");
-    //         let store = make_store(dct_50, path_50, "dct50.messagepack".into());
+        let name = match file.file_name().into_string() {
-    //         tx.send(store).expect("Thread send result error");
+            Ok(v) => v,
-    //     })
+            Err(e) => {
-    // );
+                error!("Error reading {}: {:?}:", file.path().to_string_lossy(), e);
-    // let path_30 = cfg.path.clone();
+                continue
-    // threads.push(
+            }
-    //     thread::spawn(move || {
+        };
-    //         dct_30_store = make_store(dct_30, path_30, "dct30.messagepack".into());
+        let img = match image::open(file.path()) {
-    //     })
+            Ok(v) => v,
-    // );
+            Err(e) => {
-    // let path_10 = cfg.path.clone();
+                error!("Failed to process {}: {}", name, e);
-    // threads.push(
+                continue
-    //     thread::spawn(move || {
+            }
-    //         dct_50_store = make_store(dct_10, path_10, "dct10.messagepack".into());
+        };
-    //     })
+
-    // );
+        //Store the phashes of the base image
-    for thread in threads {
+        for store in &mut stores {
-        let _ = thread.join();
+            store.store(&img, name.clone());
+        }
+
+        for mutator in &mutators {
+            let mutated = mutator.mutate(&img);
+            let mutated_name = "mut".to_string() + &mutator.tag() + &name.clone();
+            for store in &mut stores {
+                store.store(&mutated, mutated_name.clone());
+            }
+        }
     }
 
+    for store in &stores {
+        println!("{}", store);
+    }
 }

src/mutators/mod.rs

@@ -0,0 +1,81 @@
+//! Mutators that can be used to mutate an image.
+//! These mutated images can be used as a "near copy"
+
+pub trait Mutator {
+	/// Name/description of the mutator
+	fn info(&self) -> String;
+
+	/// Short tag that represents the mutator.
+	/// To be used in the filename such that it can be recognized as mutated image
+	fn tag(&self) -> String {
+		"MUT".to_string()
+	}
+
+	/// Returns the mutated form of the input image
+	fn mutate(&self, img: &image::DynamicImage) -> image::DynamicImage;
+}
+
+/// Horizontal flip/mirror
+pub struct Flip;
+impl Mutator for Flip {
+	fn info(&self) -> String {
+		"Horizontal flip".to_string()
+	}
+	fn tag(&self) -> String {
+		".flip.".to_string()
+	}
+	fn mutate(&self, img: &image::DynamicImage) -> image::DynamicImage {
+		img.fliph()
+	}
+}
+
+/// Slight hue-shift 
+pub struct Hue;
+impl Mutator for Hue {
+	fn info(&self) -> String {
+		"Hue shift".to_string()
+	}
+	fn tag(&self) -> String {
+		".hue.".to_string()
+	}
+	fn mutate(&self, img: &image::DynamicImage) -> image::DynamicImage {
+		img.huerotate(5)
+	}
+}
+
+/// Slight sharpen (unsharp mask)
+pub struct Sharp;
+impl Mutator for Sharp {
+	fn info(&self) -> String {
+		"Unsharp mask".to_string()
+	}
+	fn tag(&self) -> String {
+		".sharp.".to_string()
+	}
+	fn mutate(&self, img: &image::DynamicImage) -> image::DynamicImage {
+		img.unsharpen(1.2, 50)
+	}
+}
+
+/// Slight blur
+pub struct Blur;
+impl Mutator for Blur {
+	fn info(&self) -> String {
+		"Gaussian blur".to_string()
+	}
+	fn tag(&self) -> String {
+		".blur.".to_string()
+	}
+	fn mutate(&self, img: &image::DynamicImage) -> image::DynamicImage {
+		img.blur(1.5)
+	}
+}
+
+pub fn get_all_mutators() -> Vec<Box<dyn Mutator>> {
+	let mut mutators: Vec<Box<dyn Mutator>> = Vec::with_capacity(4);
+	mutators.push(Box::new(Flip));
+	mutators.push(Box::new(Hue));
+	mutators.push(Box::new(Sharp));
+	mutators.push(Box::new(Blur));
+	mutators
+}

src/store.rs

@@ -6,6 +6,7 @@ use crate::descriptors::Descriptor;
 use std::fs;
 use std::path::Path;
 use std::fmt;
+use image::DynamicImage;
 use log::{info, debug, error};
 use bktree::*;
 
@@ -17,8 +18,8 @@ pub enum SaveError {
 
 /// Uses a hashmap to map descriptors to buckets of files.
 /// Also keeps a BK-tree for quick distance ranking
-pub struct DescriptorStore<T: Descriptor> {
+pub struct DescriptorStore {
-	descriptor: T,
+	descriptor: Box<dyn Descriptor>,
 	/// Main hashmap that maps descriptors to buckets of filenames
 	map: HashMap<u64, Vec<String>>, 
 
@@ -32,9 +33,9 @@ pub struct DescriptorStore<T: Descriptor> {
 	seen: Option<HashSet<String>>, 
 }
 
-impl<T: Descriptor> DescriptorStore<T> {
+impl DescriptorStore {
 	/// Makes a new empty DescriptorStore with default settings
-	pub fn new(descriptor: T) -> Self {
+	pub fn new(descriptor: Box<dyn Descriptor>) -> Self {
 		let map: HashMap<u64, Vec<String>> = HashMap::new();
 		//let seen: HashSet<String> = HashSet::new();
 		//let seen = None;
@@ -154,6 +155,13 @@ impl<T: Descriptor> DescriptorStore<T> {
 		self.seen = None;
 	}
 
+	pub fn store(&mut self, img: &DynamicImage, name: String) {
+		if !self.has_value(&name) {
+			let phash = self.descriptor.describe(&img);
+			self.insert(&phash, name);
+		}
+	}
+
 	/// Nearest neighbours
 	pub fn nn(&self, from: u64, max_distance: isize) -> Vec<(&u64, isize)> {
 		let mut neighbours = self.bktree.find(from, max_distance);
@@ -224,7 +232,7 @@ impl<T: Descriptor> DescriptorStore<T> {
 	}
 }
 
-impl<T: Descriptor> fmt::Display for DescriptorStore<T> {
+impl fmt::Display for DescriptorStore {
 	fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
 		let mut output = String::new();
 		for (key, bucket) in self.map.iter() {