This module attempts to enhance contrast of a given image by equalizing its two dimensional histogram. An easy way to enhance quality of a given image is to just equalize its histogram, but despite using minimum resources and a very short process time, there are a lot of drawbacks to it. One of the ways to tackle drawbacks of histogram equalization method is to instead equalize the image's two dimensional histogram, as one dimensional histogram of a given image does not contain the image's contextual information. Tests on a multitude of images has shown, by taking contextual information of an image in addition to the image's histogram into account when attempting to enhance contrast, output images are significantly better in quality in compare to histogram equalizaion and a handful of other known methods.

You can access the article that came up with this method here.

Here is the source code for the im2dhist python library with a short description on how it's done.

Run the following to install:

pip install im2dhisteq

For images

im2dhisteq --input 'cloudy-day.jpg' --output 'assets/cloudy-day-2dhisteq.jpg' --w 6

For videos

vid2dhisteq --input 'video-input.mp4' --output 'video-output-enhanced.mp4' --w 6

The speed of the program can be significantly increased if the go library lib-im2dhist.go in src/go_libs in the package is compiled and moved to a dir in the system path. To compile, first make sure go is installed on your system. Then run

go mod init lib-im2dhist.go
go mod tidy

followed by running the script make-library.

Instead of compiling the go library lib-im2dhist.go, you can download the library based on your system configuration:

• lib-im2dhist-linux-amd64.so
• lib-im2dhist-linux-386.so
• lib-im2dhist-darwin-amd64.dylib
• lib-im2dhist-darwin-arm64.dylib

• A one minute comparative video: https://youtu.be/7LrzX2ZpLAQ
• This is a sample image and its corresponding 2d-histogram equalized image.