remove overlapped boxes

Author: xulihang

googleOCR/Utils.bas

@@ -0,0 +1,82 @@
+B4J=true
+Group=Default Group
+ModulesStructureVersion=1
+Type=StaticCode
+Version=7.8
+@EndOfDesignText@
+'Static code module
+Sub Process_Globals
+	Private fx As JFX
+End Sub
+
+
+Sub OverlappingPercent(boxGeometry1 As Map,boxGeometry2 As Map) As Double
+	'Log("boxGeometry1"&boxGeometry1)
+	'Log("boxGeometry2"&boxGeometry2)
+	Dim X1,Y1,W1,H1 As Int
+	X1=boxGeometry1.Get("X")
+	Y1=boxGeometry1.Get("Y")
+	W1=boxGeometry1.Get("width")
+	H1=boxGeometry1.Get("height")
+	Dim X2,Y2,W2,H2 As Int
+	X2=boxGeometry2.Get("X")
+	Y2=boxGeometry2.Get("Y")
+	W2=boxGeometry2.Get("width")
+	H2=boxGeometry2.Get("height")
+	Dim theSmallerX,theBiggerX,theSmallerXWidth,theBiggerXWidth As Int
+	If theSmallerOneIndex(X1,X2)=0 Then
+		theSmallerX=X1
+		theBiggerX=X2
+		theSmallerXWidth=W1
+		theBiggerXWidth=W2
+	Else
+		theSmallerX=X2
+		theBiggerX=X1
+		theSmallerXWidth=W2
+		theBiggerXWidth=W1
+	End If
+	Dim theSmallerY,theBiggerY,theSmallerYHeight,theBiggerYHeight As Int
+	If theSmallerOneIndex(Y1,Y2)=0 Then
+		theSmallerY=Y1
+		theBiggerY=Y2
+		theSmallerYHeight=H1
+		theBiggerYHeight=H2
+	Else
+		theSmallerY=Y2
+		theBiggerY=Y1
+		theSmallerYHeight=H2
+		theBiggerYHeight=H1
+	End If
+
+	If theSmallerX+theSmallerXWidth>=theBiggerX And theSmallerY+theSmallerYHeight>=theBiggerY Then
+		'overlapping
+		Dim overlappingArea As Double = (theSmallerOne(X1+W1,X2+W2)-theBiggerX)*(theSmallerOne(Y1+H1,Y2+H2)-theBiggerY)
+		Dim area1,area2 As Double
+		area1=W1*H1
+		area2=W2*H2
+		Dim theSmallArea As Int=theSmallerOne(area1,area2)
+		'Log("overlappingArea:"&overlappingArea)
+		'Log("theSmallArea:"&theSmallArea)
+		'Log("overlapping percent:")
+		'Log(overlappingArea/theSmallArea)
+		Return overlappingArea/theSmallArea
+	Else
+		Return 0
+	End If
+End Sub
+
+Sub theSmallerOneIndex(X1 As Int,X2 As Int) As Int
+	If X1<X2 Then
+		Return 0
+	Else
+		Return 1
+	End If
+End Sub
+
+Sub theSmallerOne(X1 As Int,X2 As Int) As Int
+	If X1<X2 Then
+		Return X1
+	Else
+		Return X2
+	End If
+End Sub

googleOCR/googleOCR.b4j

@@ -14,9 +14,10 @@ Library7=javaobject
 Library8=jokhttputils2
 Library9=jstringutils
 Module1=googleOCRPlugin
+Module2=Utils
 NumberOfFiles=0
 NumberOfLibraries=12
-NumberOfModules=1
+NumberOfModules=2
 Version=7.8
 @EndOfDesignText@
 #Region Project Attributes 
@@ -36,7 +37,7 @@ Sub AppStart (Form1 As Form, Args() As String)
 	MainForm.Show
     Dim t As googleOCRPlugin
 	t.Initialize
-	wait for (t.GetTextWithLocation(fx.LoadImage(File.DirApp,"1.jpg"),"zh")) Complete (done As Object)
+	wait for (t.GetTextWithLocation(fx.LoadImage(File.DirApp,"18_094.jpg"),"ja")) Complete (done As Object)
 	Log(done)
 End Sub
 

googleOCR/googleOCR.b4j.meta

@@ -1,9 +1,12 @@
 ModuleBookmarks0=
 ModuleBookmarks1=
+ModuleBookmarks2=
 ModuleBreakpoints0=
 ModuleBreakpoints1=
+ModuleBreakpoints2=
 ModuleClosedNodes0=
 ModuleClosedNodes1=
-NavigationStack=googleOCRPlugin,Initialize,7,0,googleOCRPlugin,GetNiceName,14,0,googleOCRPlugin,getLangs,39,5,googleOCRPlugin,Run,23,0,Main,AppStart,17,6,googleOCRPlugin,Paragraph2Box,332,6,googleOCRPlugin,ocr,252,6,Main,Process_Globals,5,0,googleOCRPlugin,Class_Globals,0,0
+ModuleClosedNodes2=
+NavigationStack=googleOCRPlugin,saveImgToDiskWithSizeCheck,345,0,Main,AppStart,18,6,Utils,Process_Globals,4,0,Utils,theSmallerOne,75,0,Utils,OverlappingPercent,6,0,googleOCRPlugin,Class_Globals,0,0,googleOCRPlugin,Paragraph2Box,336,6,googleOCRPlugin,GetArea,352,6,googleOCRPlugin,RemoveOverlapped,344,6,googleOCRPlugin,ocr,204,6
 SelectedBuild=0
-VisibleModules=1
+VisibleModules=1,2

googleOCR/googleOCRPlugin.bas

@@ -233,7 +233,8 @@ Sub ocr(img As B4XBitmap,lang As String) As ResumableSub
 	features.Initialize
 	Dim feature As Map
 	feature.Initialize
-	feature.Put("type","TEXT_DETECTION")
+	'feature.Put("type","TEXT_DETECTION")
+	feature.Put("type","DOCUMENT_TEXT_DETECTION")
 	features.Add(feature)
 	request.Put("image",imageMap)
 	request.Put("features",features)
@@ -280,6 +281,7 @@ Sub ocr(img As B4XBitmap,lang As String) As ResumableSub
 			Log(LastException)
 		End Try
 	End If
+	RemoveOverlapped(boxes)
 	Return boxes
 End Sub
 
@@ -339,6 +341,39 @@ Sub Paragraph2Box(paragraph As Map) As Map
 	Return box
 End Sub
 
+Sub RemoveOverlapped(boxes As List)
+	Dim new As List
+	new.Initialize
+	For i=0 To boxes.Size-1
+		Dim shouldRemove As Boolean=False
+		Dim box1 As Map=boxes.Get(i)
+		Dim geometry1 As Map
+		geometry1=box1.Get("geometry")
+		For j=0 To boxes.Size-1
+			Dim box2 As Map=boxes.Get(j)
+			Dim geometry2 As Map
+			geometry2=box2.Get("geometry")
+			If Utils.OverlappingPercent(geometry1,geometry2)>0.5 Then
+				If GetArea(geometry1)<GetArea(geometry2) Then
+					shouldRemove=True
+				End If
+			End If
+		Next
+		If shouldRemove=False Then
+			new.Add(box1)
+		End If
+	Next
+	boxes.Clear
+	boxes.Addall(new)
+End Sub
+
+Sub GetArea(boxGeometry As Map) As Int
+	Dim width,height As Int
+	width=boxGeometry.Get("width")
+	height=boxGeometry.Get("height")
+	Return width*height
+End Sub
+
 Sub getMap(key As String,parentmap As Map) As Map
 	Return parentmap.Get(key)
 End Sub