7 files changed, 289 insertions, 0 deletions
diff --git a/preproc/cmd/cleanup/main.go b/preproc/cmd/cleanup/main.go
new file mode 100644
index 0000000..7ea0c84
--- /dev/null
+++ b/preproc/cmd/cleanup/main.go
@@ -0,0 +1,62 @@
+package main
+
+// TODO: add minimum size variable (default ~30%?)
+// TODO: make into a small library
+// TODO: have the integral image specific stuff done by interface functions
+
+import (
+	"flag"
+	"fmt"
+	"image"
+	"image/draw"
+	_ "image/jpeg"
+	"image/png"
+	"log"
+	"os"
+
+	"rescribe.xyz/go.git/binarize"
+	"rescribe.xyz/go.git/preproc"
+)
+
+func main() {
+	flag.Usage = func() {
+		fmt.Fprintf(os.Stderr, "Usage: cleanup [-t thresh] [-w winsize] inimg outimg\n")
+		flag.PrintDefaults()
+	}
+	wsize := flag.Int("w", 5, "Window size for mask finding algorithm.")
+	thresh := flag.Float64("t", 0.05, "Threshold for the proportion of black pixels below which a window is determined to be the edge.")
+	flag.Parse()
+	if flag.NArg() < 2 {
+		flag.Usage()
+		os.Exit(1)
+	}
+
+	f, err := os.Open(flag.Arg(0))
+	defer f.Close()
+	if err != nil {
+		log.Fatalf("Could not open file %s: %v\n", flag.Arg(0), err)
+	}
+	img, _, err := image.Decode(f)
+	if err != nil {
+		log.Fatalf("Could not decode image: %v\n", err)
+	}
+	b := img.Bounds()
+	gray := image.NewGray(image.Rect(0, 0, b.Dx(), b.Dy()))
+	draw.Draw(gray, b, img, b.Min, draw.Src)
+
+	integral := binarize.Integralimg(gray)
+
+	lowedge, highedge := preproc.Findedges(integral, *wsize, *thresh)
+
+	clean := preproc.Wipesides(gray, lowedge, highedge)
+
+	f, err = os.Create(flag.Arg(1))
+	if err != nil {
+		log.Fatalf("Could not create file %s: %v\n", flag.Arg(1), err)
+	}
+	defer f.Close()
+	err = png.Encode(f, clean)
+	if err != nil {
+		log.Fatalf("Could not encode image: %v\n", err)
+	}
+}
diff --git a/preproc/testdata/pg1.png b/preproc/testdata/pg1.png
new file mode 100644
index 0000000..c7c4249
--- /dev/null
+++ b/preproc/testdata/pg1.png
diff --git a/preproc/testdata/pg1_integralwipesides_t0.02_w5.png b/preproc/testdata/pg1_integralwipesides_t0.02_w5.png
new file mode 100644
index 0000000..6b4ccb2
--- /dev/null
+++ b/preproc/testdata/pg1_integralwipesides_t0.02_w5.png
diff --git a/preproc/testdata/pg1_integralwipesides_t0.05_w25.png b/preproc/testdata/pg1_integralwipesides_t0.05_w25.png
new file mode 100644
index 0000000..39dc88d
--- /dev/null
+++ b/preproc/testdata/pg1_integralwipesides_t0.05_w25.png
diff --git a/preproc/testdata/pg1_integralwipesides_t0.05_w5.png b/preproc/testdata/pg1_integralwipesides_t0.05_w5.png
new file mode 100644
index 0000000..50df855
--- /dev/null
+++ b/preproc/testdata/pg1_integralwipesides_t0.05_w5.png
diff --git a/preproc/wipesides.go b/preproc/wipesides.go
new file mode 100644
index 0000000..2afe1d2
--- /dev/null
+++ b/preproc/wipesides.go
@@ -0,0 +1,122 @@
+package preproc
+
+// TODO: add minimum size variable (default ~30%?)
+// TODO: have the integral image specific stuff done by interface functions
+
+import (
+	"image"
+	"image/color"
+)
+
+type windowslice struct {
+	topleft     uint64
+	topright    uint64
+	bottomleft  uint64
+	bottomright uint64
+}
+
+func getwindowslice(i [][]uint64, x int, size int) windowslice {
+	maxy := len(i) - 1
+	maxx := x + size
+	if maxx > len(i[0])-1 {
+		maxx = len(i[0]) - 1
+	}
+
+	return windowslice{i[0][x], i[0][maxx], i[maxy][x], i[maxy][maxx]}
+}
+
+// checkwindow checks the window from x to see whether more than
+// thresh proportion of the pixels are white, if so it returns true.
+func checkwindow(integral [][]uint64, x int, size int, thresh float64) bool {
+	height := len(integral)
+	window := getwindowslice(integral, x, size)
+	// divide by 255 as each on pixel has the value of 255
+	sum := (window.bottomright + window.topleft - window.topright - window.bottomleft) / 255
+	area := size * height
+	proportion := float64(area)/float64(sum) - 1
+	return proportion <= thresh
+}
+
+// returns the proportion of the given window that is black pixels
+func proportion(integral [][]uint64, x int, size int) float64 {
+	height := len(integral)
+	window := getwindowslice(integral, x, size)
+	// divide by 255 as each on pixel has the value of 255
+	sum := (window.bottomright + window.topleft - window.topright - window.bottomleft) / 255
+	area := size * height
+	return float64(area)/float64(sum) - 1
+}
+
+// findbestedge goes through every vertical line from x to x+w to
+// find the one with the lowest proportion of black pixels.
+func findbestedge(integral [][]uint64, x int, w int) int {
+	var bestx int
+	var best float64
+
+	if w == 1 {
+		return x
+	}
+
+	right := x + w
+	for ; x < right; x++ {
+		prop := proportion(integral, x, 1)
+		if prop > best {
+			best = prop
+			bestx = x
+		}
+	}
+
+	return bestx
+}
+
+// Findedges finds the edges of the main content, by moving a window of wsize
+// from the middle of the image to the left and right, stopping when it reaches
+// a point at which there is a lower proportion of black pixels than thresh.
+func Findedges(integral [][]uint64, wsize int, thresh float64) (int, int) {
+	maxx := len(integral[0]) - 1
+	var lowedge, highedge int = 0, maxx
+
+	for x := maxx / 2; x < maxx-wsize; x++ {
+		if checkwindow(integral, x, wsize, thresh) {
+			highedge = findbestedge(integral, x, wsize)
+			break
+		}
+	}
+
+	for x := maxx / 2; x > 0; x-- {
+		if checkwindow(integral, x, wsize, thresh) {
+			lowedge = findbestedge(integral, x, wsize)
+			break
+		}
+	}
+
+	return lowedge, highedge
+}
+
+// Wipesides fills the sections of image not within the boundaries
+// of lowedge and highedge with white
+func Wipesides(img *image.Gray, lowedge int, highedge int) *image.Gray {
+	b := img.Bounds()
+	new := image.NewGray(b)
+
+	// set left edge white
+	for x := b.Min.X; x < lowedge; x++ {
+		for y := b.Min.Y; y < b.Max.Y; y++ {
+			new.SetGray(x, y, color.Gray{255})
+		}
+	}
+	// copy middle
+	for x := lowedge; x < highedge; x++ {
+		for y := b.Min.Y; y < b.Max.Y; y++ {
+			new.SetGray(x, y, img.GrayAt(x, y))
+		}
+	}
+	// set right edge white
+	for x := highedge; x < b.Max.X; x++ {
+		for y := b.Min.Y; y < b.Max.Y; y++ {
+			new.SetGray(x, y, color.Gray{255})
+		}
+	}
+
+	return new
+}
diff --git a/preproc/wipesides_test.go b/preproc/wipesides_test.go
new file mode 100644
index 0000000..76151fb
--- /dev/null
+++ b/preproc/wipesides_test.go
@@ -0,0 +1,105 @@
+package preproc
+
+// TODO: add different pages as test cases
+// TODO: test non integral img version
+
+import (
+	"flag"
+	"fmt"
+	"image"
+	"image/draw"
+	"image/png"
+	"os"
+	"testing"
+
+	"rescribe.xyz/go.git/binarize"
+)
+
+var update = flag.Bool("update", false, "update golden files")
+
+func decode(s string) (*image.Gray, error) {
+	f, err := os.Open(s)
+	defer f.Close()
+	if err != nil {
+		return nil, err
+	}
+	img, err := png.Decode(f)
+	if err != nil {
+		return nil, err
+	}
+	b := img.Bounds()
+	gray := image.NewGray(image.Rect(0, 0, b.Dx(), b.Dy()))
+	draw.Draw(gray, b, img, b.Min, draw.Src)
+	return gray, nil
+}
+
+func imgsequal(img1 *image.Gray, img2 *image.Gray) bool {
+	b := img1.Bounds()
+	if !b.Eq(img2.Bounds()) {
+		return false
+	}
+	for y := b.Min.Y; y < b.Max.Y; y++ {
+		for x := b.Min.X; x < b.Max.X; x++ {
+			r0, g0, b0, a0 := img1.At(x, y).RGBA()
+			r1, g1, b1, a1 := img2.At(x, y).RGBA()
+			if r0 != r1 {
+				return false
+			}
+			if g0 != g1 {
+				return false
+			}
+			if b0 != b1 {
+				return false
+			}
+			if a0 != a1 {
+				return false
+			}
+		}
+	}
+	return true
+}
+
+func TestWipeSides(t *testing.T) {
+	cases := []struct {
+		name   string
+		orig   string
+		golden string
+		thresh float64
+		wsize  int
+	}{
+		{"integralwipesides", "testdata/pg1.png", "testdata/pg1_integralwipesides_t0.02_w5.png", 0.02, 5},
+		{"integralwipesides", "testdata/pg1.png", "testdata/pg1_integralwipesides_t0.05_w5.png", 0.05, 5},
+		{"integralwipesides", "testdata/pg1.png", "testdata/pg1_integralwipesides_t0.05_w25.png", 0.05, 25},
+	}
+
+	for _, c := range cases {
+		t.Run(fmt.Sprintf("%s_%0.2f_%d", c.name, c.thresh, c.wsize), func(t *testing.T) {
+			var actual *image.Gray
+			orig, err := decode(c.orig)
+			if err != nil {
+				t.Fatalf("Could not open file %s: %v\n", c.orig, err)
+			}
+			integral := binarize.Integralimg(orig)
+			lowedge, highedge := Findedges(integral, c.wsize, c.thresh)
+			actual = Wipesides(orig, lowedge, highedge)
+			if *update {
+				f, err := os.Create(c.golden)
+				defer f.Close()
+				if err != nil {
+					t.Fatalf("Could not open file %s to update: %v\n", c.golden, err)
+				}
+				err = png.Encode(f, actual)
+				if err != nil {
+					t.Fatalf("Could not encode update of %s: %v\n", c.golden, err)
+				}
+			}
+			golden, err := decode(c.golden)
+			if err != nil {
+				t.Fatalf("Could not open file %s: %v\n", c.golden, err)
+			}
+			if !imgsequal(golden, actual) {
+				t.Errorf("Processed %s differs to %s\n", c.orig, c.golden)
+			}
+		})
+	}
+}