Core

Software Architecture

The imaging project highly sticks to a functional-like programming style. For this reason most of the functionality is based on Java’s Function or Consumer interface. Functions return new images and don´t touch the given object, while consumers apply changes directly onto the given object.

The core implementations are highly generic so any ImageWrapper implementation can be used with most of them. For this reason many function classes require an injected ImageFactory to create the result image. The input type can be wildcards most of the time. So if you want to use a GaussFilter which takes any input image and results in a BufferedImage wrapper use it like this:

GaussFilterFunction<?, BufferedImage> gaussFilterFunction = new GaussFilterFunction<>(ImageFactoryFactory.getImageFactory(BufferedImage.class));
ImageWrapper<BufferedImage> result = gaussFilterFunction.apply(input);

Packages

The main package is the imageprocessing package which consists different functionalities as well as submodules, as listend below.

• imageprocessing: Contains image processing functionality
  • analysis: Contains a function for calculating the ratio of a specific color
  • contour: Consists of functions of contour detection and boundary tracing
  • contrast: Consists of interfaces for contrast correction (e.q. gamma correction, histogram equalization, …)
  • conversion: Contains functions for converting an image from a given color space to another. Also contains functions for splitting/merging an image to/from channels
  • creator: Contains creator interfaces for object creation. e.q. creating a JavaImage based on given JavaLines
  • distance: Contains distance metrics and functionality for distance map calculation.
  • draw: Contains consumer implementations that draw on images
  • filter: Contains filter functionality as AnisotropicDiffusion or different convolution filters
    • highpass: Contains high pass filter e.g. for edge detection
    • lowpass: Contains low pass filter as mean or gauss filter
    • pooling: Contains pooling filters as MaxPooling
  • fitnessfunction: provides an interface for calculating fitness metrics as SumOfSquareDifferences
  • helper: Contains helper functions for normalizing an image, finding the min/max value in an image or to create a histogram for the image
  • houghspace: Contains interfaces and implementations for detecting lines using the hough space
  • interpolation: Contains interpolation functionality as BilinearInterpolation or NearestNeighbor
  • metadata: Contains functionality for extracting EXIF metadata from images
  • operator: Contains multiple base operations for images as adding two images or subtracting one image from another
  • registration: Contains an interface for registration methods
  • segmentation: Contains a segmentation function for segmenting color parts in an image.
  • transformation: Contains image transformation functions (e.g. InvertFunction, Threshold, Transformfunction, Crop, …)
  • transformers: Contains transformers for converting between different image representation (e.g. 2Byte Image, 8Byte Image, Buffered Image)
  • color: Contains transformers between different color representations (e.g. RGB to HSV)
• objectprocessing: Contains object recognition interfaces
  • compare: Contains interfaces for comparing recognized objects
  • merge: Contains interfaces for merging recognized objects
• pointprocessing: Contains functionality for processing points, pointclouds, polygons and similar point structures as convex hull calculation or boundary tracing.
  • convexhull: Contains functionality for calculating a convex hull using the graham convex hull algorithm
• storage: Contains storage services e.g. for saving/loading images or points as/from CSV files