1. Use

**W3C SVG conformance test suite**as input test files http://www.w3.org/Graphics/SVG/Test

2.

**Convert the SVG files into Bitmaps**using SharpVectorGraphics library (aka SVG#):

http://sourceforge.net/projects/svgdomcsharp

http://www.gotdotnet.com/workspaces/releases/viewuploads.aspx?id=8bcf7f55-ac4f-4778-9792-37d3e69ecf5c

3.

**Compare the Bitmaps with the expected PNG images**using FFT (Fast Fourier Transform) and MSE (Mean Square Error).

It is the Cumulative squared error between the compressed and the original image.

M N 2

MSE = 1/MN ∑ ∑ [I (x, y) –I’ (x, y)]

Y=1 X=1

I (x, y) = original image,

I’ (x, y)= approximation of decompressed image

M, N = dimensions of the images

The FFT of the images was calculated using the ExocortexDSP library.

http://sourceforge.net/projects/exocortexdsp

**More Info:**

ImageMagick also using MSE for image comparison.

http://www.imagemagick.org/script/compare.php

If MSE is not "fine" enough, we will add more compare algorithms to the process, and combine the results into a value which will estimate the similarities.

**TODO:**

* FFT of 8x8 sub regions

* have images using the same file format

* resize images to be the same size

* reduce images colors

* lowering resolution

A percentage match can be found by measuring the level of similarity of several processes:

* Color - strip boarders, compare histograms to get top colors and areas.

* Edge Detect - edge detection, and line generation, compare line lists.

* Histogram to detect shapes

## No comments:

Post a Comment