CHANGELOG.md

2.2.0

Image.hx

• Added Image.copyImageFrom

vision.ds

• Added ImageFormat.JPEG, ImageFormat.VISION

vision.tools

• Added ImageTools.loadFromFile (synchronous version)
• Added ImageTools.loadFromBytes
• Added ImageTools.loadFromURL
• Added ImageTools.exportToBytes
• Added ImageTools.exportToFile
• Deprecated ImageTools.saveToFile in favor of ImageTools.exportToFile

vision.formats

• New Subdirectory inside the vision package for cleaner image format conversions
• Added ImageIO - reads/writes images from/to different formats/frameworks
• Added support for jpeg encoding for non-js platforms using format

2.1.0

Vision.hx

• Added tint()
• Added filterForColorChannel()
• Added kmeansPosterize()
• Added kmeansGroupImageColors()
• Added simpleImageSimilarity()

Image.hx

• Added toArray()
• Fixed regression in Image.paintPixel() where alpha wasn't considered correctly

vision.ds

• Added ColorChannel.hx enum
• Added proper credit for Color.hx
• Added Color.fromFloat()
• Added Color.distanceBetween()
• Added Color.differenceBetween()
• Added Color.getAverage()
• Added ByteArray.from() Overload for Int64
• Added Matrix2D.getAverage

vision.tools

• Fixed ArrayTools min/max/average/median methods
• MathTools min/max/average/median now reference ArrayTools
• Added ArrayTools.distinct()
• Added ArrayTools.flatMap()
• Added ArrayTools.distanceTo()
• Added MathTools.SQRT2
• Added MathTools.SQRT3
• Added MathTools.toFloat for Int64

vision.algorithms

• Added KMeans class for kmeans color clustering
• Fixed Radix not supporting number types correctly
• Fixed Cramer not considering vision_quiet and added informative error messages
• Added ImageHashing class. The following hashing methods are implemented:
  • ahash
  • phash

vision.exceptions

• Added InvalidCramerSetuo exception
• Added InvalidCramerCoefficientsMatrix exception

2.0.0

Breaking!

• Typo fix: Kernal has been replaced by Kernel across the entire library

Vision.hx

• Added affineTransform()
• Added projectiveTransform()
• Added fisheyeDistortion()
• Added barrelDistortion()
• Added pincushionDistortion()
• Added smooth()
• Added posterize()
• Added sepia()
• Added pixelate()
• Added vignette()

Image.hx

• Added rotate()
• hasPixel() now receives Floats instead of Ints
• Slight optimization for getFloatingPixel(), avoids recursive call for out-of-bounds coordinates
• Fixed paintPixel and paintFloatingPixel multiplying color's alpha channels incorrectly
• Added center(), returns a point representing the center of the image
• Added pixelToRelative() & relativeToPixel(), converts between pixel and relative coordinates for an image
• Added copyViewFrom()
• Added fillEllipse()
• Fixed ellipses drawing incorrectly in some drawEllipse calls

vision.ds

• Added Matrix2D - for better representation of transformation matrices. supports:
  • getDeterminant(), getTrace()
  • invert()
  • Mathematical operations (+, -, * and /)
  • Matrix generation function (ROTATION, SHEAR, TRANSLATION, SCALE, REFLECTION, PERSPECTIVE, DEPTH, IDENTITY)
  • Pretty printing of Matrix2D
  • And more convenient functions
• Added Point3D - Representation of 3D point.
• Documented IntPoint2D and Point2D
• Added Array2D.clone()
• Added Array2D.toString()
• Added Array2D.row() and Array2D.column()
• Added haxedef vision_fancy_array_access, allowing more concise Array2D reading/writing
• Fixed critical error with Array2D get & set methods
• Added ByteArray.from, for conversion of arbitrary values to ByteArray
• Added ByteArray.resize()
• Added ByteArray.concat()
• Added ByteArray.getBytes() and ByteArray.setBytes()
• Added ByteArray.getUInt32() and ByteArray.setUInt32()
• Added ByteArray.isEmpty()
• Added ByteArray.toArray()
• Argument added to ByteArray's constructor, fillWith
• ByteArrays are now pre-filled with 0
• Added many colors to Color.hx - Off-white, Sepia, and 50+ more...
• Added haxedef to fix hl/c compilation of class Color.hx

vision.tools

• Added ArrayTools, contains several function moved from MathTools Related to Arrays.
• Added MathTools trigonometric functions using degrees:
  • sind()
  • cosd()
  • tand()
  • cotand()
  • cosecd()
  • secd()
• Added MathTools.distanceBetweenPoints() for Point3D
• Added MathTools.parseFloat() for compatibility
• Added MathTools.parseBool() for comfort :)
• Added MathTools.factorial()
• Added MathTools.gamma()
• Removed MathTools.flatten's Matrix variation, as it is no longer relevant
• Clarified parameter names in MathTools' compatibility functions
• Added return types to all MathTools functions
• Fixed ImageTools.addToScreen() not considering alpha
• Fixed canvas to image conversion not considering alpha
• Slight fix for file-name in ImageTools.saveToFile on web
• Fixed ImageTools.toShape() referencing openfl when targeting flash

vision.algorithms

• Renamed Gaussian -> Gauss
• Renamed Laplacian -> Laplace
• Added GaussJordan for matrix inversion
• Added BilinearInterpolation.interpolateMissingPixels() for filling in of missing pixels in images
• Added Cramer for solving systems of linear equations
• Fixed Radix.sort not supporting negative numbers with Int and Int64

vision.exceptions

• Added MatrixOperationError
• Added WebResponseError
• Removed double mention of pixel position in OutOfBounds exception

1.3.1

vision.algorithms

• Improved SimpleLineDetector's line filtering, reporting should be a bit better

vision.ds

• Fixed documentation on some classes
• Fixed compile error on c++, cppia & c#
• Added distanceTo() method to Point2D & IntPoint2D

1.3.0

Vision.hx

• Tweaked Vision.simpleLine2DDetection(), should report less overlapping lines

Image.hx

• Added Bitwise operators |, & and ^
• Floating pixel operations are now safe - Using out-of-bounds coordinates won't throw an error.
• Added movePixel(), allows moving a pixel from inside the image to another position, while resetting the old position. variations:
  • moveSafePixel()
  • moveFloatingPixel()
• Fixed Image.drawLine() hanging in certain scenarios
• Fixed inaccuracies in Image's view implementation, off-by-one pixels bugs should be fixed
• Fixed getSafePixel(), setSafePixel()
• Removed fromColorByteArrayAndData() Due to buggy behavior
• Added fromBytes() and toBytes() for easier cross-frameworking
• Optimized from2DArray() and to2DArray()
• Improved doc comment for private static field Image.OFFSET

vision.exceptions

• Renovated LibraryRequired to support dependencies and more accurate library inclusion methods

vision.ds

• Added Int16Point2D - An abstract over a 32-bit integer.
• Added UInt16Point2D - Same abstract, but coordinates range from 0 to 65535
• Added Color.makeRandom()
• Fixed Line2D.toString() returning each point 4 times
• Exposed Ray2D.point and updated documentation
• Added multiple get & set methods to ByteArray:
  • getUInt8()
  • setUInt8()
  • getInt8()
  • setInt8()
• Added Line2D.intersect()
• Added Line2D.distanceTo()
• Added Ray2D.distanceTo()

vision.algorithms

• Fixed Bilinear interpolation failing with some aspect ratios
• Added Gaussian.create2DKernalOfSize()
• Added Gaussian.create1DKernalOfSize()
• Added Gaussian.fastBlur, Optimized Version of Vision.gaussianBlur(). More Bare-Bones
• Deprecated Gaussian.createXxXKernal() Functions
• Deprecated Gaussian.createKernalOfSize() in favor of Gaussian.create2DKernalOfSize()

vision.tools

• Added MathTools.isBetweenRange()
• Added MathTools.getClosestPointOnRay2D()
• Added MathTools.distanceBetweenRays2D()
• Fixed MathTools.distanceFromPointOnRay2D()
• Fixed MathTools.clamp()

1.2.3

vision.ds

• Added ImageView.toString()
• Added PixelFormat - Vision now supports importing/exporting different pixel formats! supports RGB, BGR and their different alpha position variants.

vision.tools

• Implemented ImageTools.saveToFile() for JS
• Fixed ImageTools.saveToFile() saving incorrectly on sys targets
• Fixed ImageTools.loadFromFile() loading incorrectly on sys targets

1.2.2

vision.ds

• Fix DivideByZero errors on PHP when using Color.divide()
• Fix Histograms simply not working
• Optimized image.paintPixel()
• Fixed image.stamp(), should now respect alpha values.

1.2.1

Vision.hx

• Slight optimization for simpleLine2DDetection()

Image.hx

• Added from/to casting methods for JS Canvas & Image elements
• Added from/to casting methods for HaxeUI's Image element and ImageData

vision.helpers

• Added triangulation methods for drawing on images (Thanks Nanjizal)

vision.exceptions

• Fixed LibraryRequired errors being inaccurate on Ceramic

vision.tools

• Added MathTools.cropDecimal()

1.2.0

Vision.hx

• Optimized algorithm's pixel looping method
• Vision.grayscale() Now accepts an addition, optional, parameter, replacing vision_better_grayscale.

vision.ds.Image

• Image's byte offset changed from 4 to 11
• Added image.stamp() - draw an image on top of another image
• Added image.view - easy access for getting & setting the currently set image view
• Added image.forEachPixelInView() - allows iterating over only the pixels inside the current view
• Added image.hasView()
• Added image.getView()
• Added image.setView()
• Added image.removeView()
• Added image.hasPixelInView() - allows checking whether or not the given pixel is inside a given/the image's view
• Added image.setFloatingPixel()
• Added image.setSafePixel()
• Added image.paintFloatingPixel()
• Added image.paintSafePixel()
• Fixed Image.paintPixel() resetting alpha to 1
• Fixed Image.resize() not working with Nearest-Neighbor

vision.ds

• Added vision.ds.ImageView - manipulate a section of the image without having to copy/paste pixels!
• Added vision.ds.ImageViewShape
• Changed Color.grayscale()'s signature to accept an optional, extra variable. That variable enable/disables better grayscaling.
• Added Color.blackOrWhite()

1.1.1

vision.ds

• Fixed Image.underlying setter failing with vision_allow_resize

vision.tools

• Fixed forced dependency on format
• Added more documentation for ImageTools
• Fixed MathTools.clamp()
• Optimized MathTools.boundInt()

1.1.0

Vision.hx

• Added erode()
• Added dilate()
• Added combine()
• Added robertEdgeDiffOperator(), using the newly implemented RobertsCross algorithm.
• Added laplacianEdgeDiffOperator()
• Added laplacianOfGaussianEdgeDetection()
• Added bilateralDenoise(), using the newly implemented BilateralFilter algorithm.
• Added saltAndPepperNoise()
• Added dropOutNoise()
• Added whiteNoise()
• Optimized perwittEdgeDiffOperator()

vision.ds

• Added Array2D
• Added ImageFormat for image saving.
• Added WhiteNoiseRange for different white noise types (2-bit, 16-bit...)
• Added Kernal2D.LaplacianPositive
• Added Kernal2D.LaplacianNegative
• Added Color.from8Bit() to get a gray color out of a single, 8-bit color channel.
• Optimized Image.clone()

vision.tools

• Added MathTools.clamp(), which keeps a given number between a min and a max.
• Added MathTools.raise(), which raises a given 1D array/vector into a 2D array/matrix
• Added ImageTools.getNeighborsOfPixelIter(), which returns a for-loop friendly iterator over the neighboring pixels.
• Added full ImageTools.loadFromFile support on sys targets. This change allows you to load images from URLs in a cross platform fashion.
• Fixed MathTools.average, MathTools.median For C#. Fixed overload resolution failures getting thrown.

vision.exceptions

• Added LibraryRequired, thrown when a method/field depends on a missing library.
• Added ImageSavingFailed.
• Added ImageLoadingFailed

vision.helpers

• Fixed VisionThread preventing compilation on some targets. That had to do with that sys target not allowing multithreading.