Method and apparatus to locate field labels on forms
View Patent ↗Method and apparatus to identify field labels from filled forms using image processing to compare two or a few copies of the same kind of form, possibly filled out differently. Filled forms are processed to provide text strings, one for each line of text on each filled form. The text strings are converted to vectors. Vectors from different filled forms are compared to identify common words which are indicative of field labels. In an embodiment, a histogram may be generated to show frequency of occurrence of characters and words, the histogram values also being indicative of field labels.
1. A computer-implemented method comprising:
a. responsive to receiving versions of a filled form as images, processing the images to generate a text string for each line of each version;
b. for each version, generating a matrix of rows of text strings;
for each row for each version:
c. generating a best match for the row of one version with a vector within a predetermined number of rows of the vector for the row of another version;
d. constructing a character histogram for characters in the text string in that row;
e. constructing a word histogram for words in the text string in that row;
the method further comprising:
f. responsive to generating the best match, identifying field labels for the filled forms from the character histograms and the word histograms; and
g. generating a blank form with the identified field labels.
2. The computer-implemented method of claim 1 , wherein a number of said versions in a. is two or three.
3. The computer-implemented method of claim 1 , wherein the predetermined number of rows is three.
4. The computer-implemented method of claim 1 , wherein the processing comprises identifying line images in each filled form, and generating a text string from each of the identified line images.
5. The computer-implemented method of claim 1 , further comprising, responsive to generating the best match, aligning respective text lines corresponding to the vectors in the best match.
6. The computer-implemented method of claim 5 , wherein the aligning comprises scaling text in the respective text lines so that the text in the respective text lines is a same size.
7. The computer-implemented method of claim 6 , wherein the scaling comprises identifying characters in the respective text lines that are the same, and performing the scaling on the identified characters.
8. The computer-implemented method of claim 1 , wherein generating the best match comprises:
c.1. creating a vector for the text string in that row, using identified words in the text string, and assigning a number in the vector for each identified word in the text string;
c.2. comparing a vector for the row of one version with vectors within a predetermined number of rows of the vector for the row of another version;
c.3. responsive to the comparing, generating the best match.
9. The computer-implemented method of claim 8 , further comprising, for each row in each version, identifying words appearing multiple times in each of the text strings, and, for each of the words appearing multiple times, assigning the same number to each position in the vector where the word appears.
10. The computer-implemented method of claim 1 , wherein an x-axis of the word histogram represents the words in text lines and an y-axis of the word histogram represents a density of an underlying distribution of characters in the word, where a range of the density is [0, 1], using the following equation:
ρ
=
∑
i
=
1
n
f
i
mn
where ρ is the density of the underlying distribution of the characters in a word; m is a number of text lines; n is a number of characters in the word; and f i is a number count for the ith character in the word.
11. A system comprising:
a processor; and
a non-transitory memory storing instructions which, when performed by the processor, perform a method comprising:
a. responsive to receiving versions of a filled form as images, processing the images to generate a text string for each line of each version;
b. for each version, generating a matrix of rows of text strings;
for each row for each version:
c. generating a best match for the row of one version with a vector within a predetermined number of rows of the vector for the row of another version;
d. constructing a character histogram for characters in the text string in that row;
e. constructing a word histogram for words in the text string in that row;
the method further comprising:
f. responsive to generating the best match, identifying field labels for the filled forms from the character histograms and the word histograms; and
g. generating a blank form with the identified field labels.
12. The system of claim 11 , wherein a number of said versions in a. is two or three.
13. The system of claim 11 , wherein the predetermined number of rows is three.
14. The system of claim 11 , wherein the processing comprises identifying line images in each filled form, and generating a text string from each of the identified line images.
15. The system of claim 11 , wherein the method further comprises, responsive to generating the best match, aligning respective text lines corresponding to the vectors in the best match.
16. The system of claim 15 , wherein the aligning comprises scaling text in the respective text lines so that the text in the respective text lines is a same size.
17. The system of claim 16 , wherein the scaling comprises identifying characters in the respective text lines that are the same, and performing the scaling on the identified characters.
18. The system of claim 11 , wherein generating the best match comprises:
c.1. creating a vector for the text string in that row, using identified words in the text string, and assigning a number in the vector for each identified word in the text string;
c.2. comparing a vector for the row of one version with vectors within a predetermined number of rows of the vector for the row of another version;
c.3. responsive to the comparing, generating the best match.
19. The system of claim 11 , wherein the method further comprises, for each row in each version, identifying words appearing multiple times in each of the text strings, and, for each of the words appearing multiple times, assigning the same number to each position in the vector where the word appears.
20. The system of claim 11 , wherein an x-axis of the word histogram represents the words in text lines and an y-axis of the word histogram represents a density of an underlying distribution of characters in the word, where a range of the density is [0, 1], using the following equation:
ρ
=
∑
i
=
1
n
f
i
mn
where ρ is the density of the underlying distribution of the characters in a word; m is a number of text lines; n is a number of characters in the word; and f i is a number count for the ith character in the word.