public VidiReadResults.VidiReadResults[] GenerateVidiReadResults(SegmentInput Input, int StartIndex, int EndIndex, double devP)
//Generate array of ViDiReadResults
{
DoublesRemoved dblsRemoved = RemoveDoubles(Input, StartIndex, EndIndex, devP);
int n = dblsRemoved.DbleRemovedStr.Length;
VidiReadResults.VidiReadResults[] results = new VidiReadResults.VidiReadResults[n];
if (n > 0)
{
try
{
int j = 0;
while (j < n)
{
VidiReadResults.VidiReadResults res = new VidiReadResults.VidiReadResults('0', 0, 0)
{
Character = dblsRemoved.CharList[j],
Score = dblsRemoved.ScoreList[j],
Position = dblsRemoved.PosList[j]
};
results[j] = res;
j++;
}
}
catch { }
}
return(results);
}
public FinalResult SegmentAB(SegmentInput A, SegmentInput B, double devW, double devSpace)
{
FinalResult result = new FinalResult();
ResultA resA = SegmentA(A, devW);
ResultB resB = SegmentB(B, devW, devSpace);
result.Date = resA.Date;
result.DotA = resA.Dot;
result.IndexDateEnd = resA.IndexDateEnd;
result.IndexDateStart = resA.IndexDateStart;
result.DotB = resB.Dot;
result.IndexDotBStart = resB.IndexDotStart;
result.IndexDotBEnd = resB.IndexDotEnd;
return(result);
}
public ViDiReadResultList GenerateList(SegmentInput Input)
/* Generate lists of all relevant data: Score, X-position, width */
{
ViDiReadResultList result = new ViDiReadResultList();
string readResult = "";
List <double> xpos_list = new List <double>(); //list for x-positions of start of new segment
List <double> w_list = new List <double>(); //list for width of characters
List <double> score_list = new List <double>(); //list for score of characters
if (Input.Marking != null && Input.OcrString != null)
{
readResult = Input.OcrString;
IBlueMarking m = Input.Marking;
Point p = new Point();
int n = readResult.Length;
int i = 0;
while (i < n)
{
try
{
p = m.Views[0].Matches[0].Features[i].Position; //We assume only 1 model match and this model is 0
p = p.ToOriginalFromView(m.Views[0]); //Tranformation back to pixel image from Vidi image
xpos_list.Add(p.X);
score_list.Add(m.Views[0].Matches[0].Features[i].Score);
w_list.Add(m.Views[0].Matches[0].Features[i].Size.Width);
}
catch { }
}
}
else
{
readResult = "";
}
result.Read = readResult;
result.ScoreList = score_list;
result.XPosList = xpos_list;
result.WidthList = w_list;
return(result);
}
//Before segmenting we want to get rid of rubbish in the back of the string
public PreFilterResult PreFilter(SegmentInput Input)
{
PreFilterResult result = new PreFilterResult();
string str = Input.OcrString;
if (str != null)
{
int length = Input.OcrString.Length;
if (length > 0)
{
byte[] ascii = Encoding.ASCII.GetBytes(str);
char[] str_char = str.ToCharArray(); //string is read-only, have to modify the character array and convert back to string
var char_list = new List <char>(str_char); //Convert to List to be able to easily remove at position
//removing characters which not numbers or letters, start from right to left
//only remove at the end of the string
int i = length - 1;
while (i > 7) //Min. Dot length = 7
{
int a = (int)ascii[i];
if (!((a > 47 && a < 58) || (a > 64 && a < 91))) //Only capital letters and numbers 0-9 allowed
{
char_list.RemoveAt(i);
}
else
{
break; //only remove rubbisch char at end of string (, we keep order of indexes in List)
}
i--;
}
str_char = char_list.ToArray();
str = new string(str_char);
}
}
else
{
str = "";
}
result.OcrFiltered = str;
return(result);
}
// METHOD ALAIN: KORTER, EVENTUEEL UITZOEKEN... WERKT NOG NIET ivm indexes
/*
* // Splitted string at index position
*
* string temp = str;
* for (int i = 0; i < indexList.Count; i++)
* {
* temp = temp.Insert(i + indexList[i], " ");
* Debug.WriteLine(temp);
* }
* string[] splitted = temp.Trim(new char[] { ' ' }).Split(' ');
*
* // Found date . First 4 lenght string on the end and add '*' before and at the end
* int p = splitted.Length - 1;
* bool DateFound = false;
* while (p >= 0 && !DateFound)
* {
* Debug.WriteLine(splitted[p]);
* if (splitted[p].Length == 4)
* {
* result.IndexDateStart = indexList[p - 1];
* result.IndexDateEnd = indexList[p] - 1;
* splitted[p] = $"*{splitted[p]}*";
* }
* p--;
* }
*
* // Concate all to one string
* string temp2 = String.Concat(splitted);
*
* // DOT*DATE*CLUTTER => DOT,DATE,CLUTTER
* string[] Splitted2 = temp2.Split('*');
* result.Dot = Splitted2[0];
* result.Date = Splitted2[1];
* string Clutter = Splitted2[2];
*
* return result;
* }
*/
public ResultB SegmentB(SegmentInput B, double devW, double devSpace)
{
//Remove clutter from end of string of B, we presume only one segment of clutter and with a bigger space
ResultB result = new ResultB(); //final result for B
SegmentResult segmentRes = Segment(B, devW); //Segment B --> list of indexes and spaces
int nSpace = segmentRes.SpaceList.Count;
int nSegm = segmentRes.SegmentIndexList.Count;
string str = segmentRes.FilteredString;
int nStr = segmentRes.FilteredString.Length;
int indexDotEnd = 0;
if (nStr > 0) //At least 1 char in string
{
indexDotEnd = nStr - 1; //We assume last segment is not clutter or have no previous space to compare to
if (nSpace > 1) //At least 2 spaces present--> compare last space to 1st space
{
double spaceRef = segmentRes.SpaceList[0]; //take first space as reference
if (segmentRes.SpaceList[nSpace - 1] > devSpace * spaceRef) //last space larger than allowed deviation (= devSpace * spaceRef)
{
//last segment is clutter
indexDotEnd = segmentRes.SegmentIndexList[nSegm - 1] - 1;//index of last segment - 1 = end of previous segm
}
}
result.Dot = str.Substring(0, indexDotEnd + 1);
}
else //if nStr=0
{
indexDotEnd = 0;
result.Dot = "";
}
result.IndexDotStart = 0;//We presume dot always starts at 0 (no clutter in beginning)
result.IndexDotEnd = indexDotEnd;
return(result);
}
public DoublesRemoved RemoveDoubles(SegmentInput Input, int StartIndex, int EndIndex, double devP)
//Code to remove double characters at the same position based on the score
{
DoublesRemoved result = new DoublesRemoved();
IBlueMarking m = Input.Marking;
int n = (EndIndex + 1);
string ocr = Input.OcrString;
Point p = new Point();
char[] ocr_char = ocr.ToCharArray(); //string[int] is read-only--> create and modify the character array and convert back to string
List <char> char_list = new List <char>(ocr_char); //Convert to List to be able to easily remove at certain position
List <char> newChar_list = new List <char>(); //list to build new string (no doubles)
List <double> score_list = new List <double>(); //list for score (to build result array later)
List <double> pos_list = new List <double>(); //list for positions (to build result array later)
double w = 0; //Char width
double dev = devP; //consider same when difference between two positions is < dev * characterwidth
double x_curr = 0;
double x_prev = 0;
double score_prev = 0;
double score_curr = 0;
try
{
if (m != null && n > 1) //Marking present and at least 2 characters
{
int i = StartIndex; //index in Marking
int j = 0; //index for newChar_list
while (i < n) //loop trough all of characters
{
w = m.Views[0].Matches[0].Features[i].Size.Width; //Get char width
p = m.Views[0].Matches[0].Features[i].Position;
p = p.ToOriginalFromView(m.Views[0]);
x_curr = p.X;
score_curr = m.Views[0].Matches[0].Features[i].Score;
if (((x_curr - x_prev) < (dev * w)) && (j > 0)) //Compare scores when two char at one position (but not for 1st index of loop)
{
if (score_curr >= score_prev)
{
newChar_list.RemoveAt(j - 1);
score_list.RemoveAt(j - 1);
pos_list.RemoveAt(j - 1);
newChar_list.Add(char_list[j]);
score_list.Add(score_curr);
pos_list.Add(x_curr);
}
else
{
//Niks doen (niet toevoegen aan list)
}
}
else
{
newChar_list.Add(char_list[j]);
score_list.Add(score_curr);
pos_list.Add(x_curr);
}
x_prev = x_curr;
score_prev = score_curr;
i++;
j++;
}
ocr_char = newChar_list.ToArray();
result.DbleRemovedStr = ocr = new string(ocr_char);
}
else
{
result.DbleRemovedStr = "";
}
result.CharList = newChar_list;
result.PosList = pos_list;
result.ScoreList = score_list;
}
catch { result.DbleRemovedStr = ""; }
return(result);
}
public ResultA SegmentA(SegmentInput A, double devW)
{
//Segment A in Dot + Date
ResultA result = new ResultA(); //final result for A
SegmentResult segmentRes = Segment(A, devW); //Segment A --> list of indexes and spaces
int n = segmentRes.FilteredString.Length;
int indexStartDate = 0;
int indexEndDate = 0;
string str = segmentRes.FilteredString;
string date = "";
string dot = "";
List <int> indexList = new List <int>();
indexList = segmentRes.SegmentIndexList;
// METHOD ADRIAAN
//We have segmented first, now we want to check if the last segment is indeed the date timestamp
//We work from right to left!
if (n > 0 && indexList.Count > 1) //At least 2 segments
{
int i = (indexList.Count - 1); //Start from the last segment (ex. count = 12 --> last index = 11)
int j = n; //end of last segment
while (i > 0) //At least 2 segments
{
if ((j - indexList[i]) > 3) //timestamp = 4 karakters, length of segment
{
indexStartDate = indexList[i];
if (i != (indexList.Count - 1)) //if not the last segment
{
indexEndDate = indexList[i + 1] - 1; //previous index of segment -1, is end of current segment
}
else //if last segment
{
indexEndDate = n - 1;
}
break;
}
j = indexList[i]; // length of next segment = previous index - start index next segment
i--; //Work from right to left
}
}
if (n > 3 && indexList.Count == 1) //Only 1 segment and string at least 4 char --> we assume last 4 characters are timestamp
{
indexEndDate = n - 1;
indexStartDate = n - 4;
}
if (indexStartDate > 0 && indexEndDate >= (indexStartDate + 4))
{
date = str.Substring(indexStartDate, ((indexEndDate + 1) - indexStartDate));
dot = str.Substring(0, indexStartDate);
}
else
{
date = "0000";
indexEndDate = n;
dot = str;
}
result.Date = date;
result.Dot = dot;
result.IndexDateEnd = indexEndDate;
result.IndexDateStart = indexStartDate;
return(result);
}
public SegmentResult Segment(SegmentInput input, double devW)
{
//Segment into list of Segment indexes and spaces
SegmentResult result = new SegmentResult();
PreFilterResult filtered = PreFilter(input);//Filter rubbish end of string
input.OcrString = filtered.OcrFiltered;
ViDiReadResultList lists = GenerateList(input); //Use filtered string to generate lists
double w = 0;
double xCurr = 0;
double xPrev = 0;
int nChar = lists.Read.Length;
double dev = devW; //consider space when difference between two positions is > dev * characterwidth
List <double> x_list = new List <double>(); //list for width of characters
x_list = lists.XPosList;
List <double> w_list = new List <double>(); //list for width of characters
w_list = lists.WidthList;
List <double> segm_x_list = new List <double>(); //list for x-positions of start of new segment
List <int> segm_i_list = new List <int>(); //list for character indexes of start of new segment
List <double> space_list = new List <double>(); //list for size of spaces of segment
if (nChar > 0 && x_list.Count == w_list.Count && x_list.Count == nChar) //Lists have same number of elements(should be the case)
{
w = w_list.Max(); //retain widest character as reference for calculating space
//First segment starts at first character - index 0 for X-position
xCurr = x_list[0];
segm_x_list.Add(x_list[0]);
segm_i_list.Add(0);
//Loop through the remaining characters --> list of index positions with start of segments
xPrev = xCurr;
int i = 1; //start at 1, we handled 0 above
while (i < nChar)
{
xCurr = x_list[i];
if ((xCurr - xPrev) > (dev * w)) //consider space when difference between two positions is > dev * characterwidth -->New segment
{
segm_x_list.Add(xCurr);
segm_i_list.Add(i);
space_list.Add(xCurr - xPrev);
}
xPrev = xCurr;
i++;
}
}
else
{
//
}
result.Nsegments = segm_i_list.Count; //Get number of segments
result.SegmentIndexList = segm_i_list;
result.SpaceList = space_list;
result.FilteredString = lists.Read;
return(result);
}
