PixelsorterClassLib

PixelsorterClassLib is a .NET class library designed for creating advanced pixel sorting effects. It provides extensive control over the sorting criteria and directions, and it even features on-device machine learning capabilities to automatically generate masks for selective sorting.

Features

• Custom Pixel Sorting: Sort image pixels based on features like Hue, Brightness, Saturation, Lightness, Warmth, and Coolness.
• Sorting Directions: Order pixels by rows (left-to-right, right-to-left) or columns (top-to-bottom, bottom-to-top), or direct the sort into a mask.
• Machine Learning Masking: Generate binary image masks locally using ONNX models to isolate specific elements (like backgrounds or subjects) for selective sorting.
• NumSharp & ImageSharp Integration: Interacts seamlessly with pixel data as multidimensional arrays and uses ImageSharp for modern image format parsing.

Basic Usage

Below is a basic example demonstrating how to load an image, auto-generate a mask, sort the pixels by their warmth, and save the result.

using NumSharp;
using PixelsorterClassLib.Core;
using PixelsorterClassLib.Masks;

string inputPath = "input.jpg";
string outputPath = "output.jpg";

// 1. Load the image into an NDArray
NDArray imageData = Image.LoadImage(inputPath);

// 2. Get mask options

MaskOptions<BackgroundMaskOptions> options = new BackgroundMaskOptions(0.5f)

// Simpler using var
// var options = new BackgroundMask(0.5f)


// 3. Generate a binary mask using ML inference
NDArray mask = new BackgroundMask().GetMask(inputPath, options);

// 4. Sort image by warmth, left to right, within the boundaries of the mask
NDArray sortedData = Sorter.SortImage(imageData, SortBy.Warmth(), SortDirections.RowLeftToRight, mask);

// 5. Save the sorted image to disk
Image.SaveImage(sortedData, outputPath);

Available Sorting Criteria

You can easily supply your own callback function Func<Rgba32, float> or use one of the predefined sorting algorithms in the SortBy static class:

• SortBy.Hue()
• SortBy.Brightness()
• SortBy.Saturation()
• SortBy.Lightness()
• SortBy.Warmth()
• SortBy.Coolness()

You can also fetch all available sorting metrics programmatically using:

var criteria = SortBy.GetAllSortingCriteria();

Core Components

• Image: Handles reading and writing standard image formats directly to and from 3D NumSharp arrays.
• Sorter: Contains the logic to traverse pixels sequentially based on a given direction and sort segment data.
• SortBy: Supplies multiple algorithmic lens evaluations for analyzing a given pixel's characteristics.
• SortDirections: Defines the various directions in which pixels can be sorted (e.g., left-to-right, top-to-bottom).

Masks Components

• BackgroundMask: Generates a binary mask isolating the background of an image.
• CannyMask: Creates a binary mask based on edge detection using a Canny like algorithm.
• LuminanceMask Creates a mask based on the luminance of the Image.
• MaskCombiner: Combines multiple binary masks into a single mask.

License

With the commit 4e8741cf the license of this project changed from MIT to EUPL 1.2 All changes, releases etc. done before this commit stay under the MIT license.