White threshold for scanned PDFs (MuPDF)

drawcontext.h

@@ -22,6 +22,7 @@ typedef struct DrawContext {
 	int rotate;
 	double zoom;
 	double gamma;
+	unsigned char white_threshold;
 	int offset_x;
 	int offset_y;
 } DrawContext;

ffi/drawcontext.lua

@@ -11,6 +11,7 @@ typedef struct DrawContext {
 	int rotate;
 	double zoom;
 	double gamma;
+	unsigned char white_threshold;
 	int offset_x;
 	int offset_y;
 } DrawContext;
@@ -30,11 +31,13 @@ end
 function DC_mt.__index:getOffset() return self.offset_x, self.offset_y end
 function DC_mt.__index:setGamma(gamma) self.gamma = gamma end
 function DC_mt.__index:getGamma() return self.gamma end
+function DC_mt.__index:setWhiteThreshold(white_threshold) self.white_threshold = white_threshold end
+function DC_mt.__index:getWhiteThreshold() return self.white_threshold end
 
 local dctype = ffi.metatype("DrawContext", DC_mt)
 
-function DC.new(rotate, zoom, x, y, gamma)
-	return dctype(rotate or 0, zoom or 1.0, gamma or -1.0, x or 0, y or 0)
+function DC.new(rotate, zoom, x, y, gamma, white_threshold)
+	return dctype(rotate or 0, zoom or 1.0, gamma or -1.0, white_threshold or 255, x or 0, y or 0)
 end
 
 return DC

ffi/koptcontext.lua

@@ -179,6 +179,7 @@ end
 function KOPTContext_mt.__index:setTrim(trim) self.trim = trim end
 function KOPTContext_mt.__index:setWrap(wrap) self.wrap = wrap end
 function KOPTContext_mt.__index:setWhite(white) self.white = white end
+function KOPTContext_mt.__index:setPaintWhite(paint_white) self.paint_white = paint_white end
 function KOPTContext_mt.__index:setIndent(indent) self.indent = indent end
 function KOPTContext_mt.__index:setRotate(rotate) self.rotate = rotate end
 function KOPTContext_mt.__index:setColumns(columns) self.columns = columns end
@@ -599,6 +600,7 @@ function KOPTContext.new()
     kc.trim = 1
     kc.wrap = 1
     kc.white = -1
+    kc.paint_white = 0
     kc.indent = 1
     kc.rotate = 0
     kc.columns = 2
@@ -669,6 +671,7 @@ function KOPTContext.totable(kc)
     context.trim = kc.trim
     context.wrap = kc.wrap
     context.white = kc.white
+    context.paint_white = kc.paint_white
     context.indent = kc.indent
     context.rotate = kc.rotate
     context.columns = kc.columns
@@ -739,6 +742,7 @@ function KOPTContext.fromtable(context)
     kc.trim = context.trim
     kc.wrap = context.wrap
     kc.white = context.white
+    kc.paint_white = context.paint_white
     kc.indent = context.indent
     kc.rotate = context.rotate
     kc.columns = context.columns

ffi/koptcontext_h.lua

@@ -110,6 +110,7 @@ struct KOPTContext {
   int trim;
   int wrap;
   int white;
+  int paint_white;
   int indent;
   int rotate;
   int columns;

ffi/mupdf.lua

@@ -647,6 +647,29 @@ local function run_page(page, pixmap, ctm)
     M.fz_drop_device(page.ctx, dev)
     if ok == nil then merror(page.ctx, "could not run page") end
 end
+
+local function apply_white_threshold(bytes_per_pixel, samples, pixel_count, white_threshold)
+    if bytes_per_pixel == 4 then
+        for i = 0, pixel_count - 1 do
+            local idx = i * bytes_per_pixel
+            local r, g, b = samples[idx], samples[idx + 1], samples[idx + 2]
+            local brightness = 0.299 * r + 0.587 * g + 0.114 * b
+            if brightness > white_threshold then
+                samples[idx]     = 255
+                samples[idx + 1] = 255
+                samples[idx + 2] = 255
+            end
+        end
+    else
+        for i = 0, pixel_count - 1 do
+            local brightness = samples[i]
+            if brightness > white_threshold then
+                samples[i] = 255
+            end
+        end
+    end
+end
+
 --[[
 render page to blitbuffer
 
@@ -679,12 +702,16 @@ function page_mt.__index:draw_new(draw_context, width, height, offset_x, offset_
     if mupdf.bgr and self.doc.color then
         colorspace = M.fz_device_bgr(self.ctx)
     end
+    local samples = ffi.cast("unsigned char*", bb.data)
     local pix = W.mupdf_new_pixmap_with_bbox_and_data(
-        self.ctx, colorspace, bbox, nil, self.doc.color and 1 or 0, ffi.cast("unsigned char*", bb.data))
+        self.ctx, colorspace, bbox, nil, self.doc.color and 1 or 0, samples)
     if pix == nil then merror(self.ctx, "cannot allocate pixmap") end
 
     run_page(self, pix, ctm)
 
+    if draw_context.white_threshold < 255 then
+        apply_white_threshold(self.doc.color and 4 or 1, samples, width * height, draw_context.white_threshold)
+    end
     if draw_context.gamma >= 0.0 then
         M.fz_gamma_pixmap(self.ctx, pix, draw_context.gamma)
     end

thirdparty/libk2pdfopt/CMakeLists.txt

@@ -20,7 +20,7 @@ if(NOT MONOLIBTIC)
 endif()
 
 external_project(
-    DOWNLOAD GIT 59ced371378312d8f332d9a35f5b4a3c33b18954
+    DOWNLOAD GIT df18dfb0b7e33c001f1133c8b3c139b04f0e0a05
     https://github.com/koreader/libk2pdfopt.git
     PATCH_COMMAND ${PATCH_CMD}
     CMAKE_ARGS ${CMAKE_ARGS}