Skip to content

Cython implementation of 3D Region Growing algorithm

Notifications You must be signed in to change notification settings

ylmzkaan/3DRegionGrowing

Repository files navigation

3DRegionGrowing

This is an implementation of 3D Region Growing (RG) algorithm used in computer vision.

The algorithm receives a 3D image, a seed which is a point in 3D space inside the 3D image and also a 3D mask which specifies which regions of the 3D image will go through the 3D RG algorithm. If the user want the algorithm go through everywhere inside the 3D image, the input argument masks should be a 3D array of ones with a shape equal to that of 3D image's.

The user also have to specify the upper and lower threshold for the algorithm.

Flow: 1- The algorithm starts by checking all 26 Neighbors of the seed point given by the user 2- For each neighbor, if the intensity of the neighbor is within the upper and lower thresholds, the neighbor's coordinate is marked as 1 on the output mask. Finally, the neighbor is pushed into a queue. 3- When all the neighbors of the seed are finished being analysed, a new point's coordinate (newItem) is popped out of the queue. 4- The for newly popped coordinate steps 1-3 are repeated until the queue is empty.

RegionGrow3D class:

Constructer Input arguments:

images: 3D numpy array, datatype = int masks: 3D numpy array, datatype = int upperThreshold: int lowerThreshold: int

Class Methods:

outputMask = RegionGrow3D.Main(seed)

Input arguments:

seed: vector with dims=(1,3), datatype = int

Returns:

outputMask: 3D numpy array, datatype = int

About

Cython implementation of 3D Region Growing algorithm

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages