Chemical structure recognition tool

1. A Chemical Structure Recognition Tool (CSRT) to extract chemical data from an input image of a hand drawn chemical structure, said Chemical Structure Recognition Tool comprising an image scanner, and an image manipulator and analyzer coupled to the image scanner, wherein:

the image scanner is to receive a live feed of the input image from a camera and load the input image into the image manipulator and analyzer, wherein the input image is one of a photograph and a video frame of the hand drawn chemical structure sketched on a surface and captured live by the camera; and

the image manipulator and analyzer is to:

convert each color pixel of said input image into a grayscale pixel using color conversion coefficients and normalize each pixel for obtaining a grayscale image with chemically significant regions highlighted, wherein the color conversion coefficients include a red color conversion coefficient of 0.2125, a green color conversion coefficient of 0.7154, and a blue color conversion coefficient of 0.0721;

binarize said grayscale image into a binary image;

smoothen said binary image by Gaussian Smoothing;

extract chemical data from the smoothened input image by:

recognizing a shape in said binary image to be a circle to identify presence of a circle bond inside a ring;

predicting an Optical Character Recognition (OCR) region to find zones containing text;

thinning the binary image using a hit or miss morphological operation to identify specific shapes within said binary image;

detecting edges of the image by using at least one of sobel operator and canny edge detector; and

detecting a double bond and a triple bond; and

obtain output files in a digital format with the extracted chemical data from the input image.

2. The Chemical Structure Recognition Tool as claimed in claim 1 , wherein the image scanner is an image acquisition tool integrated to at least one of a digital camera, a mobile phone, a phone camera, a computer, and a scanner and wherein the image manipulator and analyzer is a software independent of type of the image scanner.

3. The Chemical Structure Recognition Tool as claimed in claim 1 , wherein said input image is accepted and output as a digital image by said image scanner.

4. A method of extracting chemical data from an input image of a hand drawn chemical structure using the Chemical Structure Recognition Tool as claimed in claim 1 , the method comprising:

receiving, by an image scanner, a live feed of the input image from a camera, wherein the input image is one of a photograph and a video frame of the hand drawn chemical structure sketched on a surface and captured live by the camera;

loading the input image by the image scanner into an image manipulator and analyzer;

converting each color pixel of said input image into a grayscale pixel using color conversion coefficients and normalizing each pixel to obtain a grayscale image with chemically significant regions highlighted, wherein the color conversion coefficients include a red color conversion coefficient of 0.2125, a green color conversion coefficient of 0.7154, and a blue color conversion coefficient of 0.0721;

binarizing said grayscale image into a binary image;

smoothing said binary image by Gaussian Smoothing;

extracting chemical data from the smoothened input image, the extracting the chemical data comprises:

recognizing a shape in said binary image to be a circle to identify presence of a circle bond inside a ring;

predicting an Optical Character Recognition (OCR) region to find zones containing text;

thinning the binary image using a hit or miss morphological operation to identify specific shapes within said binary image;

detecting edge of the image by using at least one of sobel operator and canny edge detector; and

detecting a double bond and a triple bond; and

obtaining output files in a digital format with the extracted chemical data from the input image.

5. The method as claimed in claim 4 , wherein the double bond and the triple bond are detected by using a distance formula.

6. The method as claimed in claim 4 , wherein the output files include a connection table, which identifies chemical context of texts and graphics included in the input image.

7. The method as claimed in claim 4 , wherein the shape in the binary image is recognized as the circle by determining that a calculated mean distance between the shape's edge points and an estimated circle lies in a range depending on size of the shape.

8. The method as claimed in claim 4 , wherein the digital format is one of .sdf and .mol.