image_toolbox/

lib.rs

//! Quick to start, different and random Image operations. 
//! Feel free to contribute and add new features via a Pull Request.
//! # How to use
//! In Cargo.toml 
//! ```ignore
//! [dependencies]
//! image-toolbox = "*"
//! ```
//! # The histogram struct 
//! ```
//! use image_toolbox::{Histogram, load_img};
//! use image::{DynamicImage};
//! 
//! // load img 
//! let img = load_img("./test/bright_miami.jpg").unwrap();
//! let histogram = Histogram::new(&img);
//! println!("{:?}",histogram);
//! // get the r,g,b probability of some pixel value 
//! let (p_r,p_g,p_b) : (f32,f32,f32) = histogram.probability(200);
//! ```
//! # turn a TOO bright image into normal colors
//! ```ignore
//! use image_toolbox::{load_img,normalize_brightness,save_img};
//! 
//! let img = load_img("./test/bright_miami.jpg").unwrap();
//! let new_image = normalize_brightness(&img).unwrap();
//! save_img(&new_image,"./test/result.jpg").unwrap();
//! ```

extern crate image;
extern crate math;
use std::fmt;
use image::{GenericImage, ImageBuffer,RGB,GenericImageView, DynamicImage};
use image::imageops;
use std::cmp::{min, max};

/// used to represent pixel color type in Histogram struct
#[derive(Debug, Copy, Clone)]
pub enum Pix{
    R,G,B
}

/// Histogram representation of an image 
/// represents the distribution of Red,Green,Blue pixels in an image. 
pub struct Histogram{
    r_dist : Vec<f32>,
    g_dist : Vec<f32>,
    b_dist : Vec<f32>
}
impl fmt::Debug for Histogram {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        let L = 256;
        write!(f,"\n----------------- RED -----------------\n");
        for i in 0..L{
            if self.r_dist[i] > 0.0{
                write!(f,"{} => {}\n", i, self.r_dist[i]);
            }
        }
        write!(f,"\n----------------- GREEN -----------------\n");
        for i in 0..L{
            if self.g_dist[i] > 0.0{
                write!(f,"{} => {}\n", i, self.g_dist[i]);
            }
        }
        write!(f,"\n----------------- BLUE -----------------\n");
        for i in 0..L{
            if self.b_dist[i] > 0.0{
                write!(f,"{} => {}\n", i, self.b_dist[i]);
            }
        }
        write!(f,"")
    }
}

impl Histogram{
    pub fn probability(&self,pix_val: u8)->(f32,f32,f32){
        (self.r_dist[pix_val as usize],self.g_dist[pix_val as usize],self.b_dist[pix_val as usize])
    }
    pub fn probability_of(&self,p : Pix, pix_val : u8)->f32{
        match p{
            R => self.r_dist[pix_val as usize], 
            G => self.g_dist[pix_val as usize], 
            B => self.b_dist[pix_val as usize], 
        }
    }
    pub fn new(img : & DynamicImage)->Self{
        let (width, height) = img.dimensions();
        let L = 256;
        let mut r_dist = vec![0f32;L];
        let mut g_dist = vec![0f32;L];
        let mut b_dist = vec![0f32;L];
        let sum : f32 = (width * height) as f32;
        for i in 0..width{
            for j in 0..height{
                let r_p = img.get_pixel(i,j).data[0];
                r_dist[r_p as usize] += 1.0;
                let g_p = img.get_pixel(i,j).data[1];
                g_dist[g_p as usize] += 1.0;
                let b_p = img.get_pixel(i,j).data[2];
                b_dist[b_p as usize] += 1.0;
            }
        }
        let mut sum_distros = 0.0;
        for i in 0..L{
            if r_dist[i] >= 1.0{
                r_dist[i] = r_dist[i] / sum;
                sum_distros += r_dist[i];
            }
            if g_dist[i] >= 1.0{
                g_dist[i] = g_dist[i] / sum;
            }
            if b_dist[i] >= 1.0{
                b_dist[i] = b_dist[i] / sum;
            }   
        }
        Histogram{
            r_dist: r_dist,
            g_dist: g_dist,
            b_dist: b_dist
        }
    }
}

pub fn transform_pixel(original_pixel : u8, colorType : Pix ,histogram :& Histogram)-> u8{
    let L = 256.0; 
    let new_pixel = 0; 
    let mut distros_sum = 0.0;
    let mut up_to = original_pixel as u32 +1;
    for i in 0..up_to{
        distros_sum += histogram.probability_of(colorType, i as u8);
    }
    ((L-1.0) * distros_sum) as u8
}
/// turn a very bright image into a normal looking one by using histogram equalization.
/// ```
/// use image_toolbox::{transform_from_histogram,Histogram};
/// use image::{DynamicImage};
/// 
/// // make a new 500X500 image 
/// let img = DynamicImage::new_rgba8(500,500); 
/// let histogram = Histogram::new(&img);
/// let new_image = transform_from_histogram(&img, &histogram);
/// ```
pub fn transform_from_histogram(img : &DynamicImage, hist : &Histogram)->DynamicImage{
    let (w,h) = img.dimensions();
    let mut new_img = DynamicImage::new_rgba8(w, h);
    for i in 0..w{
        for j in 0..h{
            let pixel = img.get_pixel(i, j);
            let mut r = pixel.data[0];
            let mut g = pixel.data[1];
            let mut b = pixel.data[2];
            r = transform_pixel(r, Pix::R, hist);
            g = transform_pixel(g, Pix::G, hist);
            b = transform_pixel(b, Pix::B, hist);
            let transformed_pixel = image::Rgba([r,g,b,pixel.data[3]]);
            new_img.put_pixel(i,j, transformed_pixel);
        }
    }
    new_img
}
/// Load any type of image uses `DynamicImage` type. 
/// ```
/// use image_toolbox::{load_img};
/// 
/// let img = load_img("./test/bright_miami.jpg").unwrap();
/// ```
pub fn load_img(path : &str)->Result<DynamicImage,image::ImageError>{
    let im = image::open(path)?;
    Ok(im)
}
/// Save an image to disk 
/// ```
/// use image_toolbox::{save_img};
/// use image::{DynamicImage};
/// 
/// // make a new 500X500 image 
/// let img = DynamicImage::new_rgba8(500,500); 
/// // save the image 
/// save_img(&img,"./test/empty_img.jpg").unwrap();
/// ```
pub fn save_img(img : &DynamicImage, path : &str)->Result<(),std::io::Error>{
    img.save(path)?;
    Ok(())
}
/// transform a very bright image into normal brightness based on histogram equalization
/// wrapper for `transform_from_histogram` function. 
/// ```
/// use image_toolbox::{load_img,normalize_brightness};
/// 
/// let img = load_img("./test/bright_miami.jpg").unwrap();
/// let new_image = normalize_brightness(&img).unwrap();
/// // save_img(&new_image,"./test/result.jpg").unwrap();
/// ```
pub fn normalize_brightness(img : &DynamicImage)->Result<DynamicImage,()>{
        let histogram = Histogram::new(img);
        let new_image = transform_from_histogram(&img, &histogram);
        Ok(new_image)
}

#[cfg(test)]
mod tests {
    use super::*;
    fn equals(img1 : &DynamicImage, img2 :&  DynamicImage)->bool{
        let (w1,h1) = img1.dimensions();
        let (w2,h2) = img2.dimensions();
        // compare image dimensions 
        if w1 != w2 || h1 != h2 {
            return false;
        }
        for i in 0..w1{
            for j in 0..h1{
                let p1 = img1.get_pixel(i, j);
                let r1 = p1.data[0];
                let g1 = p1.data[1];
                let b1 = p1.data[2];
                let a1 = p1.data[3];
                let p2 = img2.get_pixel(i, j);
                let r2 = p1.data[0];
                let g2 = p1.data[1];
                let b2 = p1.data[2];
                let a2 = p1.data[3];
                if r1 != r2 || g1 != g2 || b1 != b2 || a1 != a2{
                    return false;
                }
            }
        }
        true
    }
    #[test]
    fn perform_histogram_equalization() {
        let path = "./test/bright_miami.jpg";
        let img = image::open(path).unwrap();
        let histogram = Histogram::new(&img);
        let new_image = transform_from_histogram(&img, &histogram);
        // verify 
        let test_img = image::open("./test/normalized_miami.jpg").unwrap();
        assert!(equals(&test_img, &new_image),"image not equal");
    }
}