local OCR

src/local_ocr.rs

@@ -0,0 +1,109 @@
+use image::{RgbImage, DynamicImage, Rgb};
+use img_hash::{image::GenericImageView, ImageHash};
+
+use crate::image_processing;
+
+#[derive(Debug)]
+struct BoundingBox {
+    x: u32,
+    y: u32,
+    width: u32,
+    height: u32
+}
+
+fn column_has_any_dark(image: &RgbImage, x: u32) -> bool {
+    for y in 0..image.height() {
+        let [r, g, b] = image.get_pixel(x, y).0;
+        if r < 100 && g < 100 && b < 100 {
+            return true;
+        }
+    }
+    false
+}
+
+fn row_has_any_dark(image: &RgbImage, y: u32, start_x: u32, width: u32) -> bool {
+    for x in start_x..(start_x + width) {
+        let [r, g, b] = image.get_pixel(x, y).0;
+        if r < 100 && g < 100 && b < 100 {
+            return true
+        }
+    }
+    false
+}
+
+fn take_while<F: Fn(u32) -> bool>(image: &RgbImage, x: &mut u32, max: u32, f: F) {
+    while *x < max && f(*x) {
+        *x = *x + 1;
+    }
+}
+
+fn get_character_bounding_boxes(image: &RgbImage) -> Vec<BoundingBox> {
+    let mut x = 0;
+    let mut boxes = Vec::new();
+    while x < image.width() {
+        take_while(image, &mut x, image.width(), |x| !column_has_any_dark(image, x));
+
+        let start_x = x;
+        take_while(image, &mut x, image.width(), |x| column_has_any_dark(image, x));
+        let width = x - start_x;
+
+        if width > 2 {
+            let mut y = 0;
+            take_while(image, &mut y, image.height(), |y| !row_has_any_dark(image, y, start_x, width));
+
+            let start_y = y;
+
+            let mut inverse_y = 1;
+            take_while(image, &mut inverse_y, image.height(), |y| !row_has_any_dark(image, image.height() - y, start_x, width));
+            let end_y = image.height() - inverse_y - 1;
+            boxes.push(BoundingBox{
+                x,
+                y: start_y,
+                width,
+                height: end_y - start_y,
+            });
+        }
+    }
+    boxes
+}
+
+fn trim_to_bounding_box(image: &RgbImage, bounding_box: &BoundingBox) -> RgbImage {
+    let mut buffer = RgbImage::new(bounding_box.width, bounding_box.height);
+    for y in 0..bounding_box.height {
+        for x in 0..bounding_box.width {
+            buffer.put_pixel(x, y, *image.get_pixel(bounding_box.x + x, bounding_box.y + y));
+        }
+    }
+    buffer
+}
+
+fn compute_box_hashes(image: &RgbImage) -> Vec<String> {
+    let mut hashes = Vec::new();
+
+    let boxes = get_character_bounding_boxes(image);
+    for bounding_box in boxes {
+        let trimmed = trim_to_bounding_box(image, &bounding_box);
+        hashes.push(image_processing::hash_image(&trimmed))
+    }
+    hashes
+}
+
+#[test]
+fn test_bounding_boxes() {
+    let image_bytes = include_bytes!("test_data/test-image-2.png");
+    let image = image::load_from_memory(image_bytes).unwrap().to_rgb8();
+    let boxes = get_character_bounding_boxes(&image);
+    assert_eq!(boxes.len(), 10);
+    assert_ne!(boxes[0].x, 0);
+    assert_ne!(boxes[0].y, 0);
+    assert_ne!(boxes[0].height, 0);
+    assert_ne!(boxes[0].width, 0);
+}
+
+#[test]
+fn test_box_hashes() {
+    let image_bytes = include_bytes!("test_data/test-image-2.png");
+    let image = image::load_from_memory(image_bytes).unwrap().to_rgb8();
+    let hashes = compute_box_hashes(&image);
+    assert_eq!(hashes.len(), 10);
+}

src/main.rs

@@ -7,6 +7,7 @@ mod image_processing;
 mod ocr;
 mod state;
 mod stats_writer;
+mod local_ocr;
 
 use std::{
     collections::HashMap,