ANALYZING CONTENT OF DIGITAL IMAGES
Methods, apparatuses, and embodiments related to analyzing the content of digital images. A computer extracts multiple sets of visual features, which can be keypoints, based on an image of a selected object. Each of the multiple sets of visual features is extracted by a different visual feature extractor. The computer further extracts a visual word count vector based on the image of the selected object. An image query is executed based on the extracted visual features and the extracted visual word count vector to identify one or more candidate template objects of which the selected object may be an instance. When multiple candidate template objects are identified, a matching algorithm compares the selected object with the candidate template objects to determine a particular candidate template of which the selected object is an instance.
1 . A method comprising:
training a neural network with one copy of each of a set of forms;
identifying a set of closest matches between a filled-in instance of a form and the set of forms via the neural network; and
registering the filled-in instance of the form with the set of closest matches.
2 . The method of claim 1 , further comprising:
orienting the filled-in form to the set of closest matches via a Speed Up Robust Features (SURF).
3 . The method of claim 1 , further comprising:
orienting the filled-in form to the set of closest matches via a Scale Invariant Feature Transform (SIFT) with 100 key points.
4 . The method of claim 1 , wherein the set of closest matches comprises a predetermined maximum number of closest matches.
5 . The method of claim 1 , wherein the set of closest matches comprises each of the set of forms that exceed a threshold matching score with the filled-in instance of the form.
6 . The method of claim 1 , wherein the set of forms are form templates.
7 . The method of claim 1 , wherein said identifying the set of closest matches further comprises:
generating a set of bag of visual words support vectors describing graphical dimensions of each of the one copy of each of the set of forms;
generating a bag of visual words query vector describing graphical dimensions of the filled-in instance of the form; and
scoring the bag of visual words query vector against each of the set of support vectors via the neural network.
8 . The method of claim 1 , wherein the identifying is performed via an Oriented FAST Rotated BRIEF (ORB) feature detector.
9 . A system comprising:
a processor; and
a memory including a neural network trained on one copy of each of a set of forms and including instructions that when executed cause the processor to;
identify a set of closest matches between a filled-in instance of a form and the set of forms via the neural network; and
register the filled-in instance of the form with the set of closest matches.
10 . The system of claim 9 , the instructions further comprising:
orienting the filled-in form to the set of closest matches via a Speed Up Robust Features (SURF).
11 . The system of claim 9 , the instructions further comprising:
orienting the filled-in form to the set of closest matches via a Scale Invariant Feature Transform (SIFT) with 100 key points.
12 . The system of claim 9 , wherein the set of closest matches comprises a predetermined maximum number of closest matches.
13 . The system of claim 9 , wherein the set of closest matches comprises each of the set of forms that exceed a threshold matching score with the filled-in instance of the form.
14 . The system of claim 9 , wherein the set of forms are form templates.
15 . The system of claim 9 , wherein said identifying the set of closest matches further comprises:
generating a set of bag of visual words support vectors describing graphical dimensions of each of the one copy of each of the set of forms;
generating a bag of visual words query vector describing graphical dimensions of the filled-in instance of the form; and
scoring the bag of visual words query vector against each of the set of support vectors via the neural network.
16 . The system of claim 9 , wherein the identifying is performed via an Oriented FAST Rotated BRIEF (ORB) feature detector.
17 . A method comprising:
training an artificial intelligence model with one copy of each of a set of forms;
extracting a plurality of key points from a query instance of a filled in form;
identifying a set of closest matches between a filled-in instance of a form and the set of forms via a query to the artificial intelligence model including the extracted key points; and
registering the filled-in instance of the form with the set of closest matches.
18 . The method of claim 17 , wherein the set of closest matches comprises a predetermined maximum number of closest matches.
19 . The method of claim 17 , wherein the set of closest matches comprises each of the set of forms that exceed a threshold matching score with the filled-in instance of the form.
20 . The method of claim 17 , wherein the set of forms are form templates.