Initial public release

Author: kornelski

.gitignore

@@ -0,0 +1,3 @@
+dssim
+Cargo.lock
+target/

Cargo.toml

@@ -0,0 +1,30 @@
+[package]
+authors = ["Kornel <[email protected]>"]
+categories = ["multimedia::images", "command-line-utilities"]
+description = "Measures structural similarity between images using a multi-scale variant of the SSIM algorithm."
+documentation = "https://github.com/pornel/dssim#readme"
+homepage = "https://kornel.ski/dssim"
+include = ["README.md", "Cargo.toml", "tests/*.png", "src/*.rs"]
+keywords = ["ssim", "image", "comparison"]
+license = "AGPL-3.0"
+name = "dssim"
+readme = "README.md"
+repository = "https://github.com/pornel/dssim.git"
+version = "2.8.0"
+
+[[bin]]
+doctest = false
+name = "dssim"
+path = "src/main.rs"
+
+[dependencies]
+getopts = "0.2.15"
+imgref = "1.3.0"
+itertools = "0.7.2"
+lodepng = "2.0.4"
+rgb = "0.8.1"
+unzip3 = "0.1.0"
+
+[lib]
+doctest = false
+name = "dssim"

README.md

@@ -0,0 +1,73 @@
+# RGBA Structural Similarity
+
+This tool computes (dis)similarity between two or more PNG images using an algorithm approximating human vision.
+
+Comparison is done using [the SSIM algorithm](https://ece.uwaterloo.ca/~z70wang/research/ssim/) at multiple weighed resolutions.
+
+The value returned is 1/SSIM-1, where 0 means identical image, and >0 (unbounded) is amount of difference. Values are not directly comparable with other tools. [See below](#interpreting-the-values) on interpreting the values.
+
+## Features
+
+* Comparison is done in in L\*a\*b\* color space (D65 white point, sRGB gamma) with chroma subsampling. Other implementations use "RGB" or grayscale without gamma correction.
+* Supports alpha channel.
+* No OpenCV or MATLAB needed.
+   - DSSIM [version 1.x](https://github.com/pornel/dssim/tree/dssim1-c) uses C (C99) and `libpng` or Cocoa on macOS.
+   - DSSIM version 2.x is easy to build with [Rust](https://www.rust-lang.org/).
+
+## Usage
+
+    dssim file-original.png file-modified.png
+
+Will output something like "0.02341" (smaller is better) followed by a filename.
+
+You can supply multiple filenames to compare them all with the first file:
+
+    dssim file.png modified1.png modified2.png modified3.png
+
+You can save an image visualising the difference between the files:
+
+    dssim -o difference.png file.png file-modified.png
+
+The `dssim.c` file is also usable as a C library.
+
+### Interpreting the values
+
+The amount of difference goes from 0 to infinity. It's not a percentage.
+
+If you're comparing two different image compression codecs, then ensure you either:
+
+* compress images to the same file size, and then use DSSIM to compare which one is closests to the original, or
+* compress images to the same DSSIM value, and compare file sizes to see how much file size gain each option gives.
+
+[More about benchmarking image compression](https://kornel.ski/faircomparison).
+
+When you quote results, please include DSSIM version, since the scale has changed between versions.
+The version is printed when you run `dssim -h`.
+
+## Build or Download
+
+You need Rust
+
+    cargo build --release
+
+Will give you `./target/release/dssim`.
+
+## Accuracy
+
+Scores for version 2.0 measured against [TID2013][1] database:
+
+TID2013 Category | Spearman correlation
+--- | ---
+Noise  | -0.930
+Actual | -0.937
+Simple | -0.945
+Exotic | -0.842
+New    | -0.771
+Color  | -0.779
+Full   | -0.851
+
+[1]: http://www.ponomarenko.info/tid2013.htm
+
+## License
+
+DSSIM is dual-licensed under [AGPL](LICENSE) or [commercial](https://supportedsource.org/projects/dssim) license.

src/blur.rs

@@ -0,0 +1,238 @@
+const KERNEL: [f32; 9] = [
+    1./16., 2./16., 1./16.,
+    2./16., 4./16., 2./16.,
+    1./16., 2./16., 1./16.,
+];
+
+#[cfg(target_os = "macos")]
+mod mac {
+    use std;
+    use ffi::vImage_Buffer;
+    use ffi::vImageConvolve_PlanarF;
+    use ffi::vImage_Flags::kvImageEdgeExtend;
+    use super::KERNEL;
+
+    pub fn blur(src: &[f32], tmp: &mut [f32], dst: &mut [f32], width: usize, height: usize) {
+        assert_eq!(src.len(), width * height);
+        assert_eq!(dst.len(), width * height);
+
+        let srcbuf = vImage_Buffer {
+            width: width as u64,
+            height: height as u64,
+            rowBytes: width * std::mem::size_of::<f32>(),
+            data: src.as_ptr(),
+        };
+        let mut dstbuf = vImage_Buffer {
+            width: width as u64,
+            height: height as u64,
+            rowBytes: width * std::mem::size_of::<f32>(),
+            data: dst.as_mut_ptr(),
+        };
+
+        do_blur(&srcbuf, tmp, &mut dstbuf, width, height);
+    }
+
+    pub fn blur_in_place(srcdst: &mut [f32], tmp: &mut [f32], width: usize, height: usize) {
+        assert_eq!(srcdst.len(), width * height);
+
+        let srcbuf = vImage_Buffer {
+            width: width as u64,
+            height: height as u64,
+            rowBytes: width * std::mem::size_of::<f32>(),
+            data: srcdst.as_ptr(),
+        };
+        let mut dstbuf = vImage_Buffer {
+            width: width as u64,
+            height: height as u64,
+            rowBytes: width * std::mem::size_of::<f32>(),
+            data: srcdst.as_mut_ptr(),
+        };
+
+        do_blur(&srcbuf, tmp, &mut dstbuf, width, height);
+    }
+
+    pub fn do_blur(srcbuf: &vImage_Buffer<*const f32>, tmp: &mut [f32], dstbuf: &mut vImage_Buffer<*mut f32>, width: usize, height: usize) {
+        assert_eq!(tmp.len(), width * height);
+
+        unsafe {
+            let mut tmpwrbuf = vImage_Buffer {
+                width: width as u64,
+                height: height as u64,
+                rowBytes: width * std::mem::size_of::<f32>(),
+                data: tmp.as_mut_ptr(),
+            };
+            let res = vImageConvolve_PlanarF(srcbuf, &mut tmpwrbuf, std::ptr::null_mut(), 0, 0, KERNEL.as_ptr(), 3, 3, 0., kvImageEdgeExtend);
+            assert_eq!(0, res);
+
+            let tmprbuf = vImage_Buffer {
+                width: width as u64,
+                height: height as u64,
+                rowBytes: width * std::mem::size_of::<f32>(),
+                data: tmp.as_ptr(),
+            };
+            let res = vImageConvolve_PlanarF(&tmprbuf, dstbuf, std::ptr::null_mut(), 0, 0, KERNEL.as_ptr(), 3, 3, 0., kvImageEdgeExtend);
+            assert_eq!(0, res);
+        }
+    }
+}
+
+#[cfg(not(target_os = "macos"))]
+mod portable {
+    use std::cmp::min;
+    use super::KERNEL;
+
+    #[inline]
+    fn do3f(prev: &[f32], curr: &[f32], next: &[f32], i: usize) -> f32 {
+        debug_assert!(i > 0);
+
+        let c0 = i-1;
+        let c1 = i;
+        let c2 = i+1;
+
+        unsafe {
+            prev.get_unchecked(c0)*KERNEL[0] + prev.get_unchecked(c1)*KERNEL[1] + prev.get_unchecked(c2)*KERNEL[2] +
+            curr.get_unchecked(c0)*KERNEL[3] + curr.get_unchecked(c1)*KERNEL[4] + curr.get_unchecked(c2)*KERNEL[5] +
+            next.get_unchecked(c0)*KERNEL[6] + next.get_unchecked(c1)*KERNEL[7] + next.get_unchecked(c2)*KERNEL[8]
+        }
+    }
+
+    fn do3(prev: &[f32], curr: &[f32], next: &[f32], i: usize, width: usize) -> f32 {
+        let c0 = if i > 0 {i-1} else {0};
+        let c1 = i;
+        let c2 = min(i+1, width-1);
+
+        prev[c0]*KERNEL[0] + prev[c1]*KERNEL[1] + prev[c2]*KERNEL[2] +
+        curr[c0]*KERNEL[3] + curr[c1]*KERNEL[4] + curr[c2]*KERNEL[5] +
+        next[c0]*KERNEL[6] + next[c1]*KERNEL[7] + next[c2]*KERNEL[8]
+    }
+
+    pub fn blur(src: &[f32], tmp: &mut [f32], dst: &mut [f32], width: usize, height: usize) {
+        do_blur(src, tmp, width, height);
+        do_blur(tmp, dst, width, height);
+    }
+
+    pub fn do_blur(src: &[f32], dst: &mut [f32], width: usize, height: usize) {
+        assert!(width > 0);
+        assert!(width < 1<<24);
+        assert!(height > 0);
+        assert!(height < 1<<24);
+        assert!(src.len() >= width*height);
+        assert!(dst.len() >= width*height);
+
+        let mut prev = &src[0..width];
+        let mut curr = prev;
+        let mut next = prev;
+        for y in 0..height {
+            prev = curr;
+            curr = next;
+            next = if y+1 < height {&src[(y+1)*width..(y+2)*width]} else {curr};
+
+            let mut dstrow = &mut dst[y*width..y*width+width];
+
+            dstrow[0] = do3(prev, curr, next, 0, width);
+            for i in 1..width-1 {
+                dstrow[i] = do3f(prev, curr, next, i);
+            }
+            if width > 1 {
+                dstrow[width-1] = do3(prev, curr, next, width-1, width);
+            }
+        }
+    }
+
+    pub fn blur_in_place(srcdst: &mut [f32], tmp: &mut [f32], width: usize, height: usize) {
+        do_blur(srcdst, tmp, width, height);
+        do_blur(tmp, srcdst, width, height);
+    }
+}
+
+
+#[cfg(target_os = "macos")]
+pub use self::mac::*;
+
+#[cfg(not(target_os = "macos"))]
+pub use self::portable::*;
+
+#[test]
+fn blur_zero() {
+    let src = vec![0.25];
+
+    let mut tmp = vec![-55.; 1];
+    let mut dst = vec![-99.; 1];
+
+    let mut src2 = src.clone();
+
+    blur(&src[..], &mut tmp[..], &mut dst[..], 1, 1);
+    blur_in_place(&mut src2[..], &mut tmp[..], 1, 1);
+
+    assert_eq!(src2, dst);
+    assert_eq!(0.25, dst[0]);
+}
+
+#[test]
+fn blur_one() {
+    let src = vec![
+    0.,0.,0.,0.,0.,
+    0.,0.,0.,0.,0.,
+    0.,0.,1.,0.,0.,
+    0.,0.,0.,0.,0.,
+    0.,0.,0.,0.,0.,
+    ];
+    let mut tmp = vec![-55.; 5*5];
+    let mut dst = vec![999.; 5*5];
+
+    let mut src2 = src.clone();
+
+    blur(&src[..], &mut tmp[..], &mut dst[..], 5, 5);
+    blur_in_place(&mut src2[..], &mut tmp[..], 5, 5);
+
+    assert_eq!(src2, dst);
+
+    assert_eq!(1./256., dst[0]);
+    assert_eq!(1./256., dst[5*5-1]);
+    let center = 1./16.*1./16. + 2./16.*2./16. + 1./16.*1./16. +
+                 2./16.*2./16. + 4./16.*4./16. + 2./16.*2./16. +
+                 1./16.*1./16. + 2./16.*2./16. + 1./16.*1./16.;
+    assert_eq!(center, dst[2*5+2]);
+
+}
+
+#[test]
+fn blur_two() {
+    let src = vec![
+    0.,1.,1.,1.,
+    1.,1.,1.,1.,
+    1.,1.,1.,1.,
+    1.,1.,1.,1.,
+    ];
+    let mut tmp = vec![-55.; 4*4];
+    let mut dst = vec![999.; 4*4];
+
+    let mut src2 = src.clone();
+
+    blur(&src[..], &mut tmp[..], &mut dst[..], 4, 4);
+    blur_in_place(&mut src2[..], &mut tmp[..], 4, 4);
+
+    assert_eq!(src2, dst);
+
+    let z00 = 0.*1./16. + 0.*2./16. + 1.*1./16. +
+              0.*2./16. + 0.*4./16. + 1.*2./16. +
+              1.*1./16. + 1.*2./16. + 1.*1./16.;
+    let z01 =                                   0.*1./16. + 1.*2./16. + 1.*1./16. +
+                                                0.*2./16. + 1.*4./16. + 1.*2./16. +
+                                                1.*1./16. + 1.*2./16. + 1.*1./16.;
+
+    let z10 = 0.*1./16. + 0.*2./16. + 1.*1./16. +
+              1.*2./16. + 1.*4./16. + 1.*2./16. +
+              1.*1./16. + 1.*2./16. + 1.*1./16.;
+    let z11 =                                   0.*1./16. + 1.*2./16. + 1.*1./16. +
+                                                1.*2./16. + 1.*4./16. + 1.*2./16. +
+                                                1.*1./16. + 1.*2./16. + 1.*1./16.;
+    let exp = z00*1./16. + z00*2./16. + z01*1./16. +
+              z00*2./16. + z00*4./16. + z01*2./16. +
+              z10*1./16. + z10*2./16. + z11*1./16.;
+
+    assert_eq!(1., dst[3]);
+    assert_eq!(1., dst[3 * 4]);
+    assert_eq!(1., dst[4 * 4 - 1]);
+    assert_eq!(exp, dst[0]);
+}