# US 11,113,588 B2 · Randomization-based hierarchical and associatively assisted vector learning for machine vision > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,113,588 B2 - Document ID: US11113588B2 - Status: Granted - Kind: B2 - Application number: 16/590,832 - Filed: 2019-10-02 - Published: 2021-04-08 - Granted: 2021-09-07 - Earliest claimed: 2019-10-02 - Assignees: United States of America as represented by the Secretry of the Navy - Applicants: United States of America as represented by the Secretary of the Navy - Inventors: Stuart Harvey Rubin - Web page: https://patentgrep.com/patent/US11113588B2 ## CPC - G06N 20/00: Machine learning - G06F 18/214: Generating training patterns; Bootstrap methods, e.g. bagging or boosting - G06F 18/22: Matching criteria, e.g. proximity measures - G06F 18/28: Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries - G06N 5/025: Extracting rules from data - G06V 10/75: Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries - G06V 10/7715: {Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods} - G06V 10/772: Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries - G06V 10/774: Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting - G06V 30/242: Division of the character sequences into groups prior to recognition; Selection of dictionaries - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING ## Abstract An object in a digital image is automatically recognized by obtaining the digital image, generating a first vector representation of a plurality of pixels of the obtained digital image, randomizing the first vector representation to generate a first generalized vector representation of the digital image that is substantially an order of magnitude smaller than the first vector representation, and searching a repository of hierarchically generalized vector representations of previously recognized digital images, to determine whether the first generalized vector representation of the digital image matches one or more of the previously recognized digital images, within a predetermined threshold value, based on determining a focus area within the obtained digital image, using associative memory. ## Claims ### Claim 1 A method comprising: automatically recognizing an object in a digital image by: obtaining the digital image; generating a first vector representation of a plurality of pixels of the obtained digital image; randomizing the first vector representation to generate a first generalized vector representation of the digital image that is substantially an order of magnitude smaller than the first vector representation; and searching a repository of hierarchically generalized vector representations of previously recognized digital images, to determine whether the first generalized vector representation of the digital image matches one or more of the previously recognized digital images, within a predetermined threshold value, based on determining a focus area within the obtained digital image, using associative memory. ### Claim 2 (depends on 1) The method of claim 1, further comprising training instances ordered from most-general stored images to most-specific stored images, in training. ### Claim 3 (depends on 1) The method of claim 1, wherein automatically recognizing the object by generating the first vector representation of the plurality of pixels of the obtained digital image includes generating a most-specific vector representation of the plurality of pixels of the digital image. ### Claim 4 (depends on 1) The method of claim 1, wherein obtaining the digital image includes determining a focus area of a target object for obtaining the digital image. ### Claim 5 (depends on 1) The method of claim 1, wherein a camera is trained to focus attention on a subject of inference based on iterative camera movements and digital zoom. ### Claim 6 (depends on 1) The method of claim 1, wherein randomizing the first vector representation to generate the first generalized vector representation of the digital image that is substantially an order of magnitude smaller than the first vector representation includes generating the first generalized vector representation of the digital image that has a size that is substantially a power of two smaller than a size of the first vector representation. ### Claim 7 (depends on 1) The method of claim 1, further comprising iteratively randomizing the first generalized vector representation of the digital image to generate a plurality of successively more-generalized vector representations that are successively substantially an order of magnitude smaller than sequentially preceding generalized vector representations, for each of the successively more-generalized vector representations. ### Claim 8 A system comprising: at least one hardware device processor; and a non-transitory computer-readable storage medium storing instructions that are executable by the at least one hardware device processor to automatically recognize an object in a digital image by: obtaining the digital image; generating a first vector representation of a plurality of pixels of the obtained digital image; randomizing the first vector representation to generate a first generalized vector representation of the digital image that is substantially an order of magnitude smaller than the first vector representation; and searching a repository of hierarchically generalized vector representations of previously recognized digital images, to determine whether the first generalized vector representation of the digital image matches one or more of the previously recognized digital images, within a predetermined threshold value, based on determining a focus area within the obtained digital image, using associative memory. ### Claim 9 (depends on 8) The system of claim 8, wherein the instructions are executable by the at least one hardware device processor to train instances ordered from most-general stored images to most-specific stored images. ### Claim 10 (depends on 8) The system of claim 8, wherein generating the first vector representation of the plurality of pixels of the obtained digital image includes generating a most-specific vector representation of the plurality of pixels of the digital image. ### Claim 11 (depends on 8) The system of claim 8, wherein obtaining the digital image includes determining a focus area of a target object for obtaining the digital image. ### Claim 12 (depends on 8) The system of claim 8, wherein a camera is trained to focus attention on a subject of inference based on iterative camera movements and digital zoom. ### Claim 13 (depends on 8) The system of claim 8, wherein randomizing the first vector representation to generate the first generalized vector representation of the digital image that is substantially an order of magnitude smaller than the first vector representation includes generating the first generalized vector representation of the digital image that has a size that is substantially a power of two smaller than a size of the first vector representation. ### Claim 14 (depends on 8) The system of claim 8, wherein the instructions are executable by the at least one hardware device processor to iteratively randomize the first generalized vector representation of the digital image to generate a plurality of successively more-generalized vector representations that are successively substantially an order of magnitude smaller than sequentially preceding generalized vector representations, for each of the successively more-generalized vector representations. ### Claim 15 A non-transitory computer-readable storage medium storing instructions that are executable by at least one hardware device processor to automatically recognize an object in a digital image by: obtaining the digital image; generating a first vector representation of a plurality of pixels of the obtained digital image; randomizing the first vector representation to generate a first generalized vector representation of the digital image that is substantially an order of magnitude smaller than the first vector representation; and searching a repository of hierarchically generalized vector representations of previously recognized digital images, to determine whether the first generalized vector representation of the digital image matches one or more of the previously recognized digital images, within a predetermined threshold value, based on determining a focus area within the obtained digital image, using associative memory. ### Claim 16 (depends on 14) The non-transitory computer-readable storage medium of claim 14, wherein the instructions are executable by the at least one hardware device processor to train instances ordered from most-general stored images to most-specific stored images. ### Claim 17 (depends on 14) The non-transitory computer-readable storage medium of claim 14, wherein generating the first vector representation of the plurality of pixels of the obtained digital image includes generating a most-specific vector representation of the plurality of pixels of the digital image. ### Claim 18 (depends on 14) The non-transitory computer-readable storage medium of claim 14, wherein randomizing the first vector representation to generate the first generalized vector representation of the digital image that is substantially an order of magnitude smaller than the first vector representation includes generating the first generalized vector representation of the digital image that has a size that is substantially a power of two smaller than a size of the first vector representation. ### Claim 19 (depends on 14) The non-transitory computer-readable storage medium of claim 14, wherein a camera is trained to focus attention on a subject of inference based on iterative camera movements and digital zoom. ### Claim 20 (depends on 14) The non-transitory computer-readable storage medium of claim 14, wherein the instructions are executable by the at least one hardware device processor to iteratively randomize the first generalized vector representation of the digital image to generate a plurality of successively more-generalized vector representations that are successively substantially an order of magnitude smaller than sequentially preceding generalized vector representations, for each of the successively more-generalized vector representations.