Focus Stacking

This repository contains simple implementation of focus stacking technique, which is used in digital image processing. It combines multiple images taken at different focus distances to give a resulting image with a greated depth of field than any of the individual source images. [Wikipedia]

Images in example-images directory comes from http://grail.cs.washington.edu/projects/photomontage/data/bug.zip.

Requirements and Technical Info

To run this project OpenCV 3 is required (used for images disk operations).

Project was compiled and tested under Manjaro Linux 17.1.10 (kernel 4.14.47).

Build and Run Instruction

Building:

In order to build this program, please download and extract this repository and then follow below steps:

cd focus-stacking
mkdir build
cd build

cmake ../
make

Running:

After compilation two executables should be created: runAllTests and focusStacking.

In order to run tests, execute:

./runAllTests

In order to run focus stacking, execute (all arguments are optional):

./focusStacking [images_directory [bg_threshold [kernels_size [gaussian_sigma]]]]

Example:

./focusStacking ../example-images/ 70 7 5.0

After processing results are stored in the same director as executable. Results are two JPG images, i.e. focus_image.jpg and depth_map.jpg.

Parameters:

You can specify following arguments:

• images_directory [string] - relative path to directory where images are stored (default=../example-images/)
• bg_threshold [unsigned int] - threshold value used to decide whether edges are meaningful or whether they belong to image background (default=70)
• kernels_size [unsigned int] - size of gaussian and laplacian filters (default=7)
• gaussian_sigma [float] - sigma parameter of gaussian kernel (default=5.0)

Example Results

All in Focus Image	Depth Map
bg_threshold=25, kernels_size=3, gaussian_sigma=2.0
bg_threshold=70, kernels_size=7, gaussian_sigma=5.0
bg_threshold=160, kernels_size=13, gaussian_sigma=7.0

Bonus

Animated image generated with depthy.me tool.