Methods for mobile image capture of vehicle identification numbers in a non-document
Various embodiments disclosed herein are directed to methods of capturing Vehicle Identification Numbers (VIN) from images captured by a mobile device. Capturing VIN data can be useful in several applications, for example, insurance data capture applications. There are at least two types of images supported by this technology: (1) images of documents and (2) images of non-documents.
1. A method comprising using at least one hardware processor to:
receive an image from which a vehicle identification number (VIN) is to be extracted;
identify candidate text within the image by
converting the image into both a direct grayscale image, based on an assumption that the VIN is darker than a background, and an inverted grayscale image, based on an assumption that the VIN is lighter than the background,
converting the direct grayscale image and the inverted grayscale image into bi-tonal images, and
performing text segmentation on the bi-tonal images to produce the candidate text; and
perform optical character recognition (OCR) on the candidate text to identify the VIN.
2. The method of claim 1 , wherein the text segmentation comprises connected component analysis.
3. The method of claim 2 , wherein the text segmentation further comprises clustering adjacent groups of connected components from the connected component analysis into the candidate text.
4. The method of claim 1 , wherein the text segmentation comprises clustering.
5. The method of claim 1 , wherein converting the image into the direct grayscale image comprises applying a first color filter to the image, and wherein converting the image into the inverted grayscale image comprises applying a second color filter to the image.
6. The method of claim 5 , wherein the first color filter comprises:
B
(
P
)
=
R
(
P
)
*
WR
+
G
(
P
)
*
WG
+
B
(
P
)
*
WB
3
wherein B(P) is a computed brightness value of pixel P in the direct grayscale image,
wherein R(P) is a red color value of pixel P in the image,
wherein G(P) is a green color value of pixel P in the image,
wherein B(P) is a blue color value of pixel P in the image, and
wherein WR, WG, and WB are weight values.
7. The method of claim 6 , wherein the second color filter comprises:
B
(
P
)
=
(
X
-
R
(
P
)
)
*
WR
+
(
X
-
G
(
P
)
)
*
WG
+
(
X
-
B
(
P
)
)
*
WB
3
wherein X is a maximum value of pixel P.
8. The method of claim 1 , wherein performing OCR on the candidate text to identify the VIN comprises searching for one or more keywords in a result of the OCR.
9. The method of claim 8 , wherein performing OCR on the candidate text to identify the VIN further comprises performing fuzzy matching on data below or to a right of a found keyword in the result of the OCR to identify the VIN.
10. The method of claim 8 , wherein performing OCR on the candidate text to identify the VIN further comprises searching for a specific format of characters in the result of the OCR.
11. The method of claim 1 , wherein performing OCR on the candidate text to identify the VIN comprises searching for a specific format of characters in a result of the OCR.
12. The method of claim 11 , wherein the specific format is a combination of seventeen alphanumeric characters in which a last six characters are numbers.
13. The method of claim 1 , wherein receiving the image comprises receiving the image from a capture device over at least one network.
14. The method of claim 13 , further comprising using the at least one hardware processor to send the identified VIN to the capture device.
15. The method of claim 14 , wherein the capture device is a mobile phone.
16. The method of claim 1 , wherein performing optical character recognition (OCR) on the candidate text to identify the VIN comprises correcting a result of the OCR based on a check-digit calculation, wherein the check-digit calculation comprises comparing a character in a ninth position of the VIN in the result of the OCR to a modulo-11 of a sum of products that have been computed on each character in that VIN, wherein each product is computed by multiplying a respective character by a predetermined weight, and wherein the predetermined weight for the character in the ninth position is zero.
17. The method of claim 16 , wherein correcting the result of the OCR comprises replacing at least one character in the VIN in the result of the OCR with a character having a similar shape and which satisfies the check-digit calculation.
18. The method of claim 16 , wherein correcting the result of the OCR comprises using a different OCR engine to perform the OCR on the candidate text.
19. A system comprising:
at least one hardware processor; and
one or more software modules configured to, when executed by the at least one hardware processor,
receive an image from which a vehicle identification number (VIN) is to be extracted,
identify candidate text within the image by converting the image into both a direct grayscale image, based on an assumption that the VIN is darker than a background, and an inverted grayscale image, based on an assumption that the VIN is lighter than the background, converting the direct grayscale image and the inverted grayscale image into bi-tonal images, and performing text segmentation on the bi-tonal images to produce the candidate text, and
perform optical character recognition (OCR) on the candidate text to identify the VIN.
20. A non-transitory computer-readable storage medium having instructions stored thereon, wherein the instructions, when executed by a processor, cause the processor to:
receive an image from which a vehicle identification number (VIN) is to be extracted;
identify candidate text within the image by
converting the image into both a direct grayscale image, based on an assumption that the VIN is darker than a background, and an inverted grayscale image, based on an assumption that the VIN is lighter than the background,
converting the direct grayscale image and the inverted grayscale image into bi-tonal images, and
performing text segmentation on the bi-tonal images to produce the candidate text; and
perform optical character recognition (OCR) on the candidate text to identify the VIN.