diffim

Open-source, command-line, image comparing tool with masks support,
great for comparing large sets of images.

Key Features

• generates diff images to help you quickly spot the differences
• compares full images or regions of interest
• adjustable comparison threshold
• generates a full json report + a simplified csv report
• image comparison implemented using OpenCV

How to use

To start a comparison using default parameters:

> diffim.exe

The input-output folder structure is automatically created at first run. Place the images, references and masks in these folders to have them compared:

├───input\
│   ├───images
│   ├───masks
│   └───references
│   
├───output
│   ├───diff
│   ├───comparison.json
│   └───comparison.csv
│
└───diffim.exe

For every image found in references (including subfolders), a corresponding image is searched in images and the proper mask is searched in masks. If no mask is found, the full image is compared. If a folder mask (or a specific mask) is found, only the region defined by the mask is compared. The generated diff images for all images which differ are placed in the output\diff folder. The reports are stored in the output folder.

To start a comparison using the default local folders and a custom threshold of 10:

> diffim.exe 10

To start a comparison using custom paths and a custom threshold:

> diffim.exe "reference_folder" "images_folder" "masks_folder" "threshold" "OUT_report_folder" "OUT_diff_folder"

To get the latest syntax, run the help command:

> diffim.exe -help

Reports

Every comparison generates a full .json report (+ a simplified .csv one).
The report includes a comparison summary, statistics (i.e. number of identical images) and detailed data:

Compare regions of interest

To compare only parts of an image, define the regions of interest using masks and place them in the masks folder.

You can define Specific Masks, which apply to specific images (must be named as the images themselves) or you can define Folder Masks which apply to all images within a folder (except the ones which have specific masks defined) and must be named _mask.png.

The test folder includes these two types of masks. Check it out.

Installation

Install OpenCV

The image comparison algorithm is based on OpenCV. To install OpenCV:

1. download OpenCV 4.5.2 and extract it to c:\ => C:\opencv folder is created
2. add a new environmental variable OPENCV_DIR=c:\opencv\build\x64\vc15
3. add C:\opencv\build\x64\vc15\bin\ to the Path environmental variable.