Hand written word recognition using Matlab

I am working on a hand written word recognition project.

• With this project, I have collected sample hand written scripts from several different writers. Hence I have obtained different writing styles.
• For segmentation and labeling the following code is used.

I = imread('1.jpg');
bin_im = imcomplement(out);
bin_im = bwareaopen(bin_im,30);
bw1 = bin_im;

cc = bwconncomp(bw1)
cc.NumObjects

%Label the connected components
[Label,Total]=bwlabel(bw1,8);

%Rectangle containing the region
Sdata=regionprops(Label,'BoundingBox');

for i=1:Total  
%Crop all the Images 
    Img = imcrop(bw1,Sdata(i).BoundingBox);
    Name = strcat('Object Number:',num2str(i));
    pathname1 = 'C:\Users\sahanpriyanga\Documents\MATLAB\stack\Words';
    baseFileName1 = sprintf('Img%d.jpg',i);
    fullFileName1 = fullfile(pathname1,baseFileName1);
    imwrite(Img,fullFileName1);
end
This is the image that I have used.

With this code there are some problems:

• The labels are not in sequenceImage of the folder of segmented images

• The lower case letters “ i ” & “ j ” generate two figures (“ i ” gives ". " and "l" separately )
• Some letters are segmented incorrectly.

like img5 and img6 in the second figure.

The document only contains 61 characters, but after segmentation it gives 71 images.

For accurate output I need to modify the project to resolve these weaknesses and would like suggestions.

[mridul ghosh]

which algorithm(name) you followed...

[Radya Albasha]

Please I need your project

[Jagannathan M]

Unrecognized function or variable 'out'.

Error in Untitled (line 2)

bin_im = imcomplement(out);

I'm facing this error can you please help me!?

[Image Analyst]

He must have left out some code. It probably goes something like this:

I = imread('1.jpg');
out = imbinarize(I);
bin_im = imcomplement(out);

Answer 1

[Image Analyst]

See http://www.mathworks.com/help/vision/ref/ocr.html#bt548t1-2_1 for how to do OCR with MATLAB. Of course if the letters are too strange, it might not be able to identify them.

[Sahan Priyanga]

I want to recognize the hand written document which is not inline.

[Image Analyst]

See this link for algorithms: http://www.visionbib.com/bibliography/contentschar.html#OCR,%20Document%20Analysis%20and%20Character%20Recognition%20Systems

23.4.6.3 Handwriting, Cursive Script Recognition Systems

      23.4.6.3.1 Cursive Script, Word Level Recognition, Word Spotting, Language Model
      23.4.6.3.2 Cursive Script, Text Line Segmentation, Script Line, Segmentation, Text Line Extraction
      23.4.6.3.3 On-Line Cursive Script Recognition Systems 

23.4.6.4 On-Line Signatures, Online Signatures
    23.4.6.4.1 Off Line Signature Analysis
    23.4.6.4.2 Recognition of the Person from Writing, Identification, Authorship, Writer Identification
    23.4.6.4.3 Signature Recognition, Surveys, Analysis, Comparisons

Those are papers of research groups who have successfully done and published what you want to do. Needless to say, I don't have their code, so you can ask them for it, or code it up yourself from low level MATLAB functions. Good luck.

Answer 2

[Iftahatus Shofwan]

How i can fix this problem, about “ i ” & “ j ” generate two figures (“ i ” gives ". " and "l" separately ). "i" become one figure only. please give me references. thanks

[Image Analyst]

I thought I did, here in this link.

I'm sure people have worked on and handled cases where the letter is in multiple parts before.