9884534579
For each row the convolution kernel is evaluated at fixed positions around "u". By calculating these values once for each row, a huge speedup is achieved. |
||
---|---|---|
.travis.yml | ||
converter.go | ||
filters.go | ||
LICENSE | ||
README.md | ||
resize_test.go | ||
resize.go | ||
sinc_test.go | ||
sinc.go |
Resize
Image resizing for the Go programming language with common interpolation methods.
Installation
$ go get github.com/nfnt/resize
It's that easy!
Usage
Import package with
import "github.com/nfnt/resize"
Resize creates a scaled image with new dimensions (width
, height
) using the interpolation function interp
.
If either width
or height
is set to 0, it will be set to an aspect ratio preserving value.
resize.Resize(width, height uint, img image.Image, interp resize.InterpolationFunction) image.Image
The provided interpolation functions are (from fast to slow execution time)
NearestNeighbor
: Nearest-neighbor interpolationBilinear
: Bilinear interpolationBicubic
: Bicubic interpolationMitchellNetravali
: Mitchell-Netravali interpolationLanczos2Lut
: Lanczos resampling with a=2 using a look-up table for fast computationLanczos2
: Lanczos resampling with a=2Lanczos3Lut
: Lanczos resampling with a=3 using a look-up table for fast computationLanczos3
: Lanczos resampling with a=3
Which of these methods gives the best results depends on your use case.
Sample usage:
package main
import (
"github.com/nfnt/resize"
"image/jpeg"
"log"
"os"
)
func main() {
// open "test.jpg"
file, err := os.Open("test.jpg")
if err != nil {
log.Fatal(err)
}
// decode jpeg into image.Image
img, err := jpeg.Decode(file)
if err != nil {
log.Fatal(err)
}
file.Close()
// resize to width 1000 using Lanczos resampling
// and preserve aspect ratio
m := resize.Resize(1000, 0, img, resize.Lanczos3)
out, err := os.Create("test_resized.jpg")
if err != nil {
log.Fatal(err)
}
defer out.Close()
// write new image to file
jpeg.Encode(out, m, nil)
}
Downsizing Samples
Downsizing is not as simple as it might look like. Images have to be filtered before they are scaled down, otherwise aliasing might occur. Filtering is highly subjective: Applying too much will blur the whole image, too little will make aliasing become apparent. Resize tries to provide sane defaults that should suffice in most cases.
Artificial sample
Nearest-Neighbor |
Bilinear |
---|---|
Bicubic |
Mitchell-Netravali |
Lanczos2 |
Lanczos3 |
Real-Life sample
Nearest-Neighbor |
Bilinear |
---|---|
Bicubic |
Mitchell-Netravali |
Lanczos2 |
Lanczos3 |
License
Copyright (c) 2012 Jan Schlicht janschlicht@gmail.com Resize is released under a MIT style license.