func TestResize(t *testing.T) {
var img0, img1 *image.Gray
// Testing various sizes and parallelization settings
w, h := 10, 20
img0 = image.NewGray(image.Rect(0, 0, w, h))
sz := []struct{ w0, h0, w1, h1 int }{
{w, h, w, h},
{w * 2, h, w * 2, h},
{w, h * 2, w, h * 2},
{w * 2, h * 2, w * 2, h * 2},
{w / 2, h, w / 2, h},
{w, 0, w, h},
{0, h, w, h},
{w * 2, 0, w * 2, h * 2},
{0, h / 2, w / 2, h / 2},
{0, 0, 0, 0},
{1, -1, 0, 0},
{-1, 1, 0, 0},
}
rfilters := []Resampling{
NearestNeighborResampling,
BoxResampling,
LinearResampling,
CubicResampling,
LanczosResampling,
}
for _, prlz := range []bool{true, false} {
for _, z := range sz {
for _, f := range rfilters {
g := New(Resize(z.w0, z.h0, f))
g.SetParallelization(prlz)
img1 := image.NewGray(g.Bounds(img0.Bounds()))
g.Draw(img1, img0)
w2, h2 := img1.Bounds().Dx(), img1.Bounds().Dy()
if w2 != z.w1 || h2 != z.h1 {
t.Errorf("resize %s %dx%d: expected %dx%d got %dx%d", f, z.w0, z.h0, z.w1, z.h1, w2, h2)
}
}
}
}
// Nearest filter resize
img0 = image.NewGray(image.Rect(-1, -1, 4, 1))
img0.Pix = []uint8{
1, 2, 3, 4, 5,
6, 7, 8, 0, 1,
}
img1Exp := image.NewGray(image.Rect(0, 0, 2, 2))
img1Exp.Pix = []uint8{
2, 4,
7, 0,
}
f := Resize(2, 2, NearestNeighborResampling)
img1 = image.NewGray(f.Bounds(img0.Bounds()))
f.Draw(img1, img0, nil)
if img1.Bounds().Size() != img1Exp.Bounds().Size() {
t.Errorf("expected %v got %v", img1Exp.Bounds().Size(), img1.Bounds().Size())
}
if !comparePix(img1Exp.Pix, img1.Pix) {
t.Errorf("expected %v got %v", img1Exp.Pix, img1.Pix)
}
// Box Filter resize
img0 = image.NewGray(image.Rect(-1, -1, 3, 1))
img0.Pix = []uint8{
1, 2, 2, 1,
4, 5, 8, 9,
}
img1Exp = image.NewGray(image.Rect(0, 0, 2, 1))
img1Exp.Pix = []uint8{
3, 5,
}
f = Resize(2, 1, BoxResampling)
img1 = image.NewGray(f.Bounds(img0.Bounds()))
f.Draw(img1, img0, nil)
if img1.Bounds().Size() != img1Exp.Bounds().Size() {
t.Errorf("expected %v got %v", img1Exp.Bounds().Size(), img1.Bounds().Size())
}
if !comparePix(img1Exp.Pix, img1.Pix) {
t.Errorf("expected %v got %v", img1Exp.Pix, img1.Pix)
}
// Empty image should remain empty and not panic
img0 = &image.Gray{}
f = Resize(100, 100, BoxResampling)
img1 = image.NewGray(f.Bounds(img0.Bounds()))
f.Draw(img1, img0, nil)
if img1.Bounds().Dx() != 0 || img1.Bounds().Dy() != 0 {
t.Errorf("empty image resized is not empty: %dx%d", img1.Bounds().Dx(), img1.Bounds().Dy())
}
// Testing kernel values outside the window
for _, f := range rfilters {
if f.Kernel(f.Support()+0.000001) != 0 {
t.Errorf("filter %s value outside support != 0", f)
}
}
// Testing spline and sinc edge cases
if sinc(0) != 1 {
t.Errorf("sinc(0) != 1")
}
if bcspline(-2.0, 0.0, 0.5) != 0 {
t.Errorf("bcspline(2.0, ...) != 0")
}
if (resamp{name: "test"}).String() != "test" {
t.Error("resamplingStruct String() fail")
}
testData := []struct {
desc string
w, h int
r Resampling
srcb, dstb image.Rectangle
srcPix, dstPix []uint8
}{
{
"resize (1, 2 -> 1, 1; box; non-alpha)",
1, 1, BoxResampling,
image.Rect(0, 0, 1, 2),
image.Rect(0, 0, 1, 1),
[]uint8{
0xff, 0x00, 0x00, 0xff,
0x00, 0xff, 0x00, 0xff,
},
[]uint8{
0x80, 0x80, 0x00, 0xff,
},
},
{
"resize (1, 2 -> 1, 1; box; alpha)",
1, 1, BoxResampling,
image.Rect(0, 0, 1, 2),
image.Rect(0, 0, 1, 1),
[]uint8{
0xff, 0x00, 0x00, 0xff,
0x00, 0xff, 0x00, 0x00,
},
[]uint8{
0xff, 0x00, 0x00, 0x80,
},
},
{
"resize (1, 2 -> 1, 3; linear; alpha)",
1, 3, LinearResampling,
image.Rect(0, 0, 1, 2),
image.Rect(0, 0, 1, 3),
[]uint8{
0xff, 0x00, 0x00, 0xff,
0x00, 0xff, 0x00, 0x00,
},
[]uint8{
0xff, 0x00, 0x00, 0xff,
0xff, 0x00, 0x00, 0x80,
0x00, 0x00, 0x00, 0x00,
},
},
}
for _, d := range testData {
src := image.NewNRGBA(d.srcb)
src.Pix = d.srcPix
f := Resize(d.w, d.h, d.r)
dst := image.NewNRGBA(f.Bounds(src.Bounds()))
f.Draw(dst, src, nil)
if !checkBoundsAndPix(dst.Bounds(), d.dstb, dst.Pix, d.dstPix) {
t.Errorf("test [%s] failed: %#v, %#v", d.desc, dst.Bounds(), dst.Pix)
}
}
}
func TestResize(t *testing.T) {
var img0, img1 *image.Gray
// Testing various sizes and parallelization settings
w, h := 10, 20
img0 = image.NewGray(image.Rect(0, 0, w, h))
sz := []struct{ w0, h0, w1, h1 int }{
{w, h, w, h},
{w * 2, h, w * 2, h},
{w, h * 2, w, h * 2},
{w * 2, h * 2, w * 2, h * 2},
{w / 2, h, w / 2, h},
{w, 0, w, h},
{0, h, w, h},
{w * 2, 0, w * 2, h * 2},
{0, h / 2, w / 2, h / 2},
{0, 0, 0, 0},
{1, -1, 0, 0},
{-1, 1, 0, 0},
}
rfilters := []Resampling{
NearestNeighborResampling,
BoxResampling,
LinearResampling,
CubicResampling,
LanczosResampling,
}
for _, prlz := range []bool{true, false} {
for _, z := range sz {
for _, f := range rfilters {
g := New(Resize(z.w0, z.h0, f))
g.SetParallelization(prlz)
img1 := image.NewGray(g.Bounds(img0.Bounds()))
g.Draw(img1, img0)
w2, h2 := img1.Bounds().Dx(), img1.Bounds().Dy()
if w2 != z.w1 || h2 != z.h1 {
t.Errorf("resize %s %dx%d: expected %dx%d got %dx%d", f, z.w0, z.h0, z.w1, z.h1, w2, h2)
}
}
}
}
// Nearest filter resize
img0 = image.NewGray(image.Rect(-1, -1, 4, 1))
img0.Pix = []uint8{
1, 2, 3, 4, 5,
6, 7, 8, 0, 1,
}
img1_exp := image.NewGray(image.Rect(0, 0, 2, 2))
img1_exp.Pix = []uint8{
2, 4,
7, 0,
}
f := Resize(2, 2, NearestNeighborResampling)
img1 = image.NewGray(f.Bounds(img0.Bounds()))
f.Draw(img1, img0, nil)
if img1.Bounds().Size() != img1_exp.Bounds().Size() {
t.Errorf("expected %v got %v", img1_exp.Bounds().Size(), img1.Bounds().Size())
}
if !comparePix(img1_exp.Pix, img1.Pix) {
t.Errorf("expected %v got %v", img1_exp.Pix, img1.Pix)
}
// Box Filter resize
img0 = image.NewGray(image.Rect(-1, -1, 3, 1))
img0.Pix = []uint8{
1, 2, 2, 1,
4, 5, 8, 9,
}
img1_exp = image.NewGray(image.Rect(0, 0, 2, 1))
img1_exp.Pix = []uint8{
3, 5,
}
f = Resize(2, 1, BoxResampling)
img1 = image.NewGray(f.Bounds(img0.Bounds()))
f.Draw(img1, img0, nil)
if img1.Bounds().Size() != img1_exp.Bounds().Size() {
t.Errorf("expected %v got %v", img1_exp.Bounds().Size(), img1.Bounds().Size())
}
if !comparePix(img1_exp.Pix, img1.Pix) {
t.Errorf("expected %v got %v", img1_exp.Pix, img1.Pix)
}
// Empty image should remain empty and not panic
img0 = &image.Gray{}
f = Resize(100, 100, BoxResampling)
img1 = image.NewGray(f.Bounds(img0.Bounds()))
f.Draw(img1, img0, nil)
if img1.Bounds().Dx() != 0 || img1.Bounds().Dy() != 0 {
t.Errorf("empty image resized is not empty: %dx%d", img1.Bounds().Dx(), img1.Bounds().Dy())
}
// Testing kernel values outside the window
for _, f := range rfilters {
if f.Kernel(f.Support()+0.000001) != 0 {
t.Errorf("filter %s value outside support != 0", f)
}
}
// Testing spline and sinc edge cases
if sinc(0) != 1 {
t.Errorf("sinc(0) != 1")
}
if bcspline(-2.0, 0.0, 0.5) != 0 {
t.Errorf("bcspline(2.0, ...) != 0")
}
if (resamplingStruct{name: "test"}).String() != "test" {
t.Error("resamplingStruct String() fail")
}
}
