moparisthebest
/
Simba
mirror of https://github.com/moparisthebest/Simba
1
0
Fork 0
Code Issues Releases Wiki Activity

More OCR changes, but mainly OCRBench changes. 

git-svn-id: http://www.villavu.com/repositories/merlijn/mufasa@314 3f818213-9676-44b0-a9b4-5e4c4e03d09d
This commit is contained in:
Wizzup? 2009-12-22 01:19:29 +00:00
parent 19abae25e1
commit 80ce9744c6
5 changed files with 259 additions and 99 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

159
Projects/OCRBench/project1.lpi
View File

@ -7,7 +7,7 @@
<TargetFileExt Value=""/>
<Icon Value="0"/>
<UseXPManifest Value="True"/>
<ActiveEditorIndexAtStart Value="0"/>
<ActiveEditorIndexAtStart Value="3"/>
</General>
<VersionInfo>
<ProjectVersion Value=""/>
@ -31,12 +31,12 @@
<PackageName Value="LCL"/>
</Item1>
</RequiredPackages>
<Units Count="9">
<Units Count="11">
<Unit0>
<Filename Value="project1.lpr"/>
<IsPartOfProject Value="True"/>
<UnitName Value="project1"/>
<UsageCount Value="24"/>
<UsageCount Value="27"/>
</Unit0>
<Unit1>
<Filename Value="unit1.pas"/>
@ -44,10 +44,10 @@
<ComponentName Value="Form1"/>
<ResourceBaseClass Value="Form"/>
<UnitName Value="Unit1"/>
<CursorPos X="95" Y="72"/>
<TopLine Value="44"/>
<CursorPos X="64" Y="131"/>
<TopLine Value="102"/>
<EditorIndex Value="0"/>
<UsageCount Value="24"/>
<UsageCount Value="27"/>
<Loaded Value="True"/>
</Unit1>
<Unit2>
@ -58,9 +58,7 @@
<UnitName Value="colourhistory"/>
<CursorPos X="10" Y="457"/>
<TopLine Value="437"/>
<EditorIndex Value="5"/>
<UsageCount Value="12"/>
<Loaded Value="True"/>
<UsageCount Value="13"/>
</Unit2>
<Unit3>
<Filename Value="../SAMufasaGUI/testunit.pas"/>
@ -78,25 +76,25 @@
<CursorPos X="21" Y="52"/>
<TopLine Value="32"/>
<EditorIndex Value="2"/>
<UsageCount Value="12"/>
<UsageCount Value="14"/>
<Loaded Value="True"/>
</Unit4>
<Unit5>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<UnitName Value="ocr"/>
<CursorPos X="10" Y="290"/>
<TopLine Value="313"/>
<CursorPos X="86" Y="342"/>
<TopLine Value="315"/>
<EditorIndex Value="3"/>
<UsageCount Value="12"/>
<UsageCount Value="14"/>
<Loaded Value="True"/>
</Unit5>
<Unit6>
<Filename Value="../../Units/MMLCore/bitmaps.pas"/>
<UnitName Value="bitmaps"/>
<CursorPos X="49" Y="368"/>
<TopLine Value="337"/>
<EditorIndex Value="4"/>
<UsageCount Value="11"/>
<CursorPos X="25" Y="436"/>
<TopLine Value="416"/>
<EditorIndex Value="6"/>
<UsageCount Value="13"/>
<Loaded Value="True"/>
</Unit6>
<Unit7>
@ -105,7 +103,7 @@
<CursorPos X="58" Y="184"/>
<TopLine Value="160"/>
<EditorIndex Value="1"/>
<UsageCount Value="11"/>
<UsageCount Value="13"/>
<Loaded Value="True"/>
</Unit7>
<Unit8>
@ -114,127 +112,145 @@
<TopLine Value="1205"/>
<UsageCount Value="10"/>
</Unit8>
<Unit9>
<Filename Value="../../Units/MMLAddon/tpa.pas"/>
<UnitName Value="tpa"/>
<CursorPos X="68" Y="798"/>
<TopLine Value="779"/>
<EditorIndex Value="5"/>
<UsageCount Value="11"/>
<Loaded Value="True"/>
</Unit9>
<Unit10>
<Filename Value="../../Units/MMLCore/mufasatypesutil.pas"/>
<UnitName Value="mufasatypesutil"/>
<CursorPos X="1" Y="1"/>
<TopLine Value="1"/>
<EditorIndex Value="4"/>
<UsageCount Value="10"/>
<Loaded Value="True"/>
</Unit10>
</Units>
<JumpHistory Count="30" HistoryIndex="29">
<Position1>
<Filename Value="unit1.pas"/>
<Caret Line="65" Column="43" TopLine="33"/>
<Caret Line="87" Column="27" TopLine="56"/>
</Position1>
<Position2>
<Filename Value="unit1.pas"/>
<Caret Line="55" Column="70" TopLine="22"/>
<Caret Line="72" Column="4" TopLine="56"/>
</Position2>
<Position3>
<Filename Value="unit1.pas"/>
<Caret Line="54" Column="70" TopLine="22"/>
<Caret Line="70" Column="70" TopLine="56"/>
</Position3>
<Position4>
<Filename Value="unit1.pas"/>
<Caret Line="30" Column="60" TopLine="22"/>
<Filename Value="../../Units/MMLCore/ocrutil.pas"/>
<Caret Line="169" Column="23" TopLine="156"/>
</Position4>
<Position5>
<Filename Value="unit1.pas"/>
<Caret Line="31" Column="60" TopLine="22"/>
<Caret Line="71" Column="29" TopLine="44"/>
</Position5>
<Position6>
<Filename Value="unit1.pas"/>
<Caret Line="32" Column="60" TopLine="22"/>
<Filename Value="../../Units/MMLCore/ocrutil.pas"/>
<Caret Line="184" Column="58" TopLine="160"/>
</Position6>
<Position7>
<Filename Value="unit1.pas"/>
<Caret Line="31" Column="60" TopLine="22"/>
<Caret Line="74" Column="107" TopLine="58"/>
</Position7>
<Position8>
<Filename Value="unit1.pas"/>
<Caret Line="33" Column="60" TopLine="22"/>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="57" Column="33" TopLine="37"/>
</Position8>
<Position9>
<Filename Value="unit1.pas"/>
<Caret Line="84" Column="50" TopLine="44"/>
<Filename Value="../../Units/MMLAddon/tpa.pas"/>
<Caret Line="30" Column="64" TopLine="25"/>
</Position9>
<Position10>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="81" Column="22" TopLine="60"/>
<Filename Value="../../Units/MMLAddon/tpa.pas"/>
<Caret Line="63" Column="20" TopLine="44"/>
</Position10>
<Position11>
<Filename Value="unit1.pas"/>
<Caret Line="75" Column="33" TopLine="49"/>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="89" Column="20" TopLine="70"/>
</Position11>
<Position12>
<Filename Value="unit1.pas"/>
<Caret Line="79" Column="15" TopLine="42"/>
<Caret Line="94" Column="57" TopLine="71"/>
</Position12>
<Position13>
<Filename Value="../../Units/MMLCore/bitmaps.pas"/>
<Caret Line="98" Column="21" TopLine="75"/>
<Filename Value="unit1.pas"/>
<Caret Line="96" Column="57" TopLine="71"/>
</Position13>
<Position14>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="31" Column="54" TopLine="25"/>
<Filename Value="unit1.pas"/>
<Caret Line="98" Column="61" TopLine="82"/>
</Position14>
<Position15>
<Filename Value="../../Units/MMLCore/ocrutil.pas"/>
<Caret Line="180" Column="111" TopLine="157"/>
<Filename Value="unit1.pas"/>
<Caret Line="114" Column="13" TopLine="88"/>
</Position15>
<Position16>
<Filename Value="../../Units/MMLCore/ocrutil.pas"/>
<Caret Line="181" Column="13" TopLine="160"/>
<Filename Value="unit1.pas"/>
<Caret Line="96" Column="61" TopLine="77"/>
</Position16>
<Position17>
<Filename Value="../../Units/MMLCore/ocrutil.pas"/>
<Caret Line="57" Column="51" TopLine="36"/>
<Filename Value="unit1.pas"/>
<Caret Line="103" Column="10" TopLine="77"/>
</Position17>
<Position18>
<Filename Value="unit1.pas"/>
<Caret Line="72" Column="83" TopLine="57"/>
<Caret Line="102" Column="8" TopLine="84"/>
</Position18>
<Position19>
<Filename Value="unit1.pas"/>
<Caret Line="78" Column="38" TopLine="65"/>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="42" Column="36" TopLine="25"/>
</Position19>
<Position20>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="42" Column="34" TopLine="25"/>
<Filename Value="unit1.pas"/>
<Caret Line="108" Column="47" TopLine="84"/>
</Position20>
<Position21>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="78" Column="70" TopLine="62"/>
<Caret Line="333" Column="23" TopLine="301"/>
</Position21>
<Position22>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="290" Column="10" TopLine="265"/>
<Filename Value="unit1.pas"/>
<Caret Line="109" Column="28" TopLine="89"/>
</Position22>
<Position23>
<Filename Value="unit1.pas"/>
<Caret Line="75" Column="17" TopLine="46"/>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="356" Column="43" TopLine="325"/>
</Position23>
<Position24>
<Filename Value="unit1.pas"/>
<Caret Line="76" Column="21" TopLine="56"/>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="343" Column="17" TopLine="324"/>
</Position24>
<Position25>
<Filename Value="unit1.pas"/>
<Caret Line="87" Column="27" TopLine="56"/>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="346" Column="11" TopLine="331"/>
</Position25>
<Position26>
<Filename Value="unit1.pas"/>
<Caret Line="72" Column="4" TopLine="56"/>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="355" Column="37" TopLine="331"/>
</Position26>
<Position27>
<Filename Value="unit1.pas"/>
<Caret Line="70" Column="70" TopLine="56"/>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="351" Column="33" TopLine="332"/>
</Position27>
<Position28>
<Filename Value="../../Units/MMLCore/ocrutil.pas"/>
<Caret Line="169" Column="23" TopLine="156"/>
<Filename Value="../../Units/MMLCore/bitmaps.pas"/>
<Caret Line="368" Column="49" TopLine="337"/>
</Position28>
<Position29>
<Filename Value="unit1.pas"/>
<Caret Line="71" Column="29" TopLine="44"/>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="351" Column="37" TopLine="332"/>
</Position29>
<Position30>
<Filename Value="../../Units/MMLCore/ocrutil.pas"/>
<Caret Line="184" Column="58" TopLine="160"/>
<Filename Value="../../Units/MMLCore/ocr.pas"/>
<Caret Line="377" Column="57" TopLine="347"/>
</Position30>
</JumpHistory>
</ProjectOptions>
@ -256,7 +272,8 @@
</Options>
</Linking>
<Other>
<CustomOptions Value="-dOCRDEBUG"/>
<CustomOptions Value="-dOCRDEBUG
-dOCRTPA"/>
<CompilerPath Value="$(CompPath)"/>
</Other>
</CompilerOptions>

35
Projects/OCRBench/unit1.lfm
View File

@ -1,12 +1,13 @@
object Form1: TForm1
Left = 493
Height = 305
Top = 293
Left = 462
Height = 306
Top = 264
Width = 609
ActiveControl = BitmapButton
Caption = 'Form1'
ClientHeight = 305
ClientHeight = 306
ClientWidth = 609
OnCreate = FormCreate
LCLVersion = '0.9.29'
object OCRButton: TButton
Left = 512
@ -47,15 +48,33 @@ object Form1: TForm1
Top = 112
Width = 79
Caption = 'Shadow?'
OnChange = FShadowChange
TabOrder = 3
end
object SplitEdit: TEdit
Left = 512
Height = 27
Top = 168
Width = 88
TabOrder = 4
Text = '1'
end
object SplitLabel: TLabel
Left = 512
Height = 18
Top = 144
Width = 88
AutoSize = False
Caption = 'Split Spacing:'
ParentColor = False
end
object OCRFileOpen: TOpenDialog
Filter = '.bmp'
left = 528
top = 192
left = 416
top = 277
end
object UpCharsDialog: TSelectDirectoryDialog
left = 528
top = 160
left = 448
top = 277
end
end

46
Projects/OCRBench/unit1.pas
View File

@ -14,6 +14,8 @@ type
TForm1 = class(TForm)
BitmapButton: TButton;
SplitLabel: TLabel;
SplitEdit: TEdit;
FShadow: TCheckBox;
PathButton: TButton;
OCRButton: TButton;
@ -21,6 +23,8 @@ type
OCRFileOpen: TOpenDialog;
UpCharsDialog: TSelectDirectoryDialog;
procedure BitmapButtonClick(Sender: TObject);
procedure FormCreate(Sender: TObject);
procedure FShadowChange(Sender: TObject);
procedure OCRButtonClick(Sender: TObject);
procedure PathButtonClick(Sender: TObject);
private
@ -48,6 +52,7 @@ Var
x,y: integer;
s: string;
Shadow: boolean;
Spacing: Integer;
begin
if not FileExists(BitmapPath) then
@ -70,6 +75,7 @@ begin
Form1.Image1.Canvas.Brush.Color := 0;
Form1.Image1.Canvas.Rectangle(0, 0, Form1.Image1.Canvas.Width, Form1.Image1.Canvas.Height);
// create and init client
C := TClient.Create;
bmp := TMufasaBitmap.Create;
bmp.LoadFromFile(BitmapPath);
@ -77,15 +83,39 @@ begin
Shadow :=FShadow.Checked;
try
Spacing := StrToInt(Form1.SplitEdit.Text);
except
if shadow then
begin
MessageBox(0,pchar('Spacing could not be parsed.' +
'Defaulting to 2' ), Pchar('Space Error'), MB_OK);
Spacing := 2;
end
else
begin
MessageBox(0,pchar('Spacing could not be parsed.' +
'Defaulting to 1' ), Pchar('Space Error'), MB_OK);
Spacing := 1;
end;
end;
writeln('Spacing: ' + Inttostr(spacing));
// DS + .. + DS because InitOCR wants the directory of the Fonts, not UpChars
// only.
C.MOCR.InitTOCR(UpTextPath + DS + '..' + DS, Shadow);
s := C.MOCR.GetUpTextAt(7,7, Shadow);
s := C.MOCR.GetUpTextAtEx(7,7, Shadow, Spacing);
// write to debugbmp
for y := 0 to C.MOCR.debugbmp.Height - 1 do
for x := 0 to C.MOCR.debugbmp.Width -1 do
Form1.Image1.Canvas.Pixels[x,y] := C.MOCR.debugbmp.FastGetPixel(x,y);
// print ocr'ed text
Form1.Image1.Canvas.Font.Color:=clRed;
Form1.Image1.Canvas.TextOut(0, C.MOCR.debugbmp.Height, s);
Form1.Image1.Picture.SaveToFile('/tmp/ocrbench.bmp');
C.Free;
end;
@ -95,6 +125,20 @@ begin
BitmapPath := OCRFileOpen.FileName;
end;
procedure TForm1.FormCreate(Sender: TObject);
begin
{BitmapPath := '/home/merlijn/Programs/mufasa/pics/uptext2.bmp';
UpTextPath := '/home/merlijn/Programs/mufasa/Fonts/UpChars';}
end;
procedure TForm1.FShadowChange(Sender: TObject);
begin
if Form1.FShadow.Checked then
Form1.SplitEdit.Text:='2'
else
Form1.SplitEdit.Text:='1';
end;
procedure TForm1.PathButtonClick(Sender: TObject);
begin
if UpCharsDialog.Execute then

3
Units/MMLAddon/mmlpsthread.pas
View File

@ -98,7 +98,7 @@ threadvar
CurrThread : TMMLPSThread;
implementation
uses
dtmutil,
colour_conv,dtmutil,
{$ifdef mswindows}windows,{$endif}
uPSC_std, uPSC_controls,uPSC_classes,uPSC_graphics,uPSC_stdctrls,uPSC_forms,
uPSC_extctrls, //Compile-libs
@ -108,7 +108,6 @@ uses
Graphics, //For Graphics types
math, //Maths!
strutils,
colour_conv,
input,
tpa, //Tpa stuff
forms,//Forms

115
Units/MMLCore/ocr.pas
View File

@ -38,7 +38,8 @@ uses
constructor Create(Owner: TObject);
destructor Destroy; override;
function InitTOCR(path: string; shadow: Boolean): boolean;
function getTextPointsIn(sx, sy, w, h: Integer; shadow: boolean): TNormArray;
function getTextPointsIn(sx, sy, w, h: Integer; shadow: boolean; spacing: Integer): TNormArray;
function GetUpTextAtEx(atX, atY: integer; shadow: boolean; Spacing: Integer): string;
function GetUpTextAt(atX, atY: integer; shadow: boolean): string;
private
Client: TObject;
@ -54,7 +55,7 @@ uses
implementation
uses
colour_conv, client, files;
colour_conv, client, files, tpa, mufasatypesutil;
const
ocr_Limit_High = 190;
@ -75,7 +76,7 @@ const
Non optimised. ;-)
}
function TMOCR.getTextPointsIn(sx, sy, w, h: Integer; shadow: boolean): TNormArray;
function TMOCR.getTextPointsIn(sx, sy, w, h: Integer; shadow: boolean; spacing: integer): TNormArray;
var
bmp: TMufasaBitmap;
x,y: integer;
@ -84,7 +85,12 @@ var
{$IFDEF OCRDEBUG}
dx,dy: integer;
{$ENDIF}
{$IFDEF OCRTPA}
t: tpointarray;
at, atf,att: T2DPointArray;
pc: integer;
max_len: integer;
{$ENDIF}
begin
bmp := TMufasaBitmap.Create;
@ -97,7 +103,7 @@ begin
{$IFDEF OCRDEBUG}
debugbmp := TMufasaBitmap.Create;
debugbmp.SetSize(w, (h + 2) * 4);
debugbmp.SetSize(w, (h + 2) * 5);
{$ENDIF}
{$IFDEF OCRSAVEBITMAP}
@ -266,11 +272,6 @@ begin
for y := 0 to bmp.Height - 1 do
for x := 0 to bmp.Width - 1 do
begin
{ if bmp.fastgetpixel(x,y) <> clPurple then
begin
bmp.FastSetPixel(x,y,0);
continue;
end; }
if bmp.fastgetpixel(x,y) = clPurple then
begin
bmp.FastSetPixel(x,y,0);
@ -292,17 +293,33 @@ begin
for y := 0 to bmp.Height -1 do
bmp.fastsetpixel(0, y, 0);
{$IFDEF OCRTPA}
pc := 0;
setlength(t, bmp.Height * bmp.Width);
{$ENDIF}
setlength(n, bmp.Height * bmp.Width);
for y := 0 to bmp.Height - 1 do
for x := 0 to bmp.Width - 1 do
begin
if bmp.fastgetpixel(x,y) > 0 then
n[x + y * bmp.width] := 1
begin
n[x + y * bmp.width] := 1;
{$IFDEF OCRTPA}
t[pc] := point(x,y);
inc(pc);
{$ENDIF}
end
else
n[x + y * bmp.width] := 0;
n[x
+ y * bmp.width] := 0;
end;
{$IFDEF OCRTPA}
setlength(t,pc);
{$ENDIF}
result := n;
{$IFDEF OCRSAVEBITMAP}
bmp.SaveToFile('/tmp/ocrfinal.bmp');
@ -312,6 +329,64 @@ begin
for dx := 0 to bmp.width - 1 do
debugbmp.fastsetpixel(dx,dy+h+h+h,bmp.fastgetpixel(dx,dy));
{$ENDIF}
{$IFDEF OCRTPA}
at:=splittpaex(t,spacing,bmp.height);
{
// this was to split extra large points into smaller ones, but it usually won't help
if shadow then
max_len := 30
else
max_len := 50;
for x := 0 to high(at) do
begin
if length(at[x]) > max_len then
begin
setlength(t,0);
// t := at[x];
att := splittpaex(at[x], 1, bmp.height);
for y := 0 to high(att) do
begin
setlength(atf,length(atf)+1);
atf[high(atf)] := convtpaarr(att[y]);
end;
end else
begin
setlength(atf,length(atf)+1);
atf[high(atf)] := convtpaarr(at[x]);
end;
end;
for x := 0 to high(atf) do
begin
pc := random(clWhite);
for y := 0 to high(atf[x]) do
bmp.FastSetPixel(atf[x][y].x, atf[x][y].y, pc);
end; }
for x := 0 to high(at) do
begin
if length(at[x]) > 70 then
begin
for y := 0 to high(at[x]) do
bmp.FastSetPixel(at[x][y].x, at[x][y].y, clOlive);
end else
begin
pc := random(clWhite);
for y := 0 to high(at[x]) do
bmp.FastSetPixel(at[x][y].x, at[x][y].y, pc);
end;
end;
{$IFDEF OCRDEBUG}
for dy := 0 to bmp.height - 1 do
for dx := 0 to bmp.width - 1 do
debugbmp.fastsetpixel(dx,dy+h+h+h+h,bmp.fastgetpixel(dx,dy));
{$ENDIF}
{$ENDIF}
bmp.Free;
{ Dangerous removes all pixels that had no pixels on x-1 or x+1}
{ for y := 0 to bmp.Height - 2 do
@ -369,19 +444,25 @@ begin
result := false;
end;
function TMOCR.GetUpTextAt(atX, atY: integer; shadow: boolean): string;
function TMOCR.GetUpTextAtEx(atX, atY: integer; shadow: boolean; spacing: Integer): string;
var
n:Tnormarray;
ww, hh: integer;
begin
ww := 400;
hh := 20;
n := getTextPointsIn(atX, atY, ww, hh, shadow);
n := getTextPointsIn(atX, atY, ww, hh, shadow, spacing);
Result := ocrDetect(n, ww, hh, OCRData[0]);
//writeln(result);
end;
function TMOCR.GetUpTextAt(atX, atY: integer; shadow: boolean): string;
begin
if shadow then
result := GetUpTextAtEx(atX, atY, shadow, 2)
else
result := GetUpTextAtEx(atX, atY, shadow, 1);
end;
{
function TMOCR.GetUpTextAt(atX, atY: integer): string;