# US 12,192,648 B2 · Image processing device, method for training machine learning model, identification device, and image processing method > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,192,648 B2 - Document ID: US12192648B2 - Status: Granted - Kind: B2 - Application number: 18/109,351 - Filed: 2023-02-14 - Published: 2023-06-22 - Granted: 2025-01-07 - Earliest claimed: 2020-08-27 - Assignees: Panasonic Intellectual Property Corporation of America - Applicants: Panasonic Intellectual Property Corporation of America - Inventors: Satoshi Sato; Yasunori Ishii; Sotaro Tsukizawa; Changxin Zhou; Takrit Tanasnitikul; Pongsak Lasang - Web page: https://patentgrep.com/patent/US12192648B2 ## CPC - H04N 23/80: Camera processing pipelines; Components thereof - G06F 21/6254: {by anonymising data, e.g. decorrelating personal data from the owner's identification} - G06V 10/774: Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting - G06V 10/82: using neural networks - H04N 23/55: Optical parts specially adapted for electronic image sensors; Mounting thereof - H04N 23/617: Upgrading or updating of programs or applications for camera control - H04N 23/95: Computational photography systems, e.g. light-field imaging systems - H04N 23/955: for lensless imaging - H04N 25/60: Noise processing, e.g. detecting, correcting, reducing or removing noise - G06T 7/00: Image analysis - G06V 40/172: {Classification, e.g. identification} - H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract An image processing device includes a noise adder that obtains a captured image from an image capturing device including a mask having at least one aperture, an MPH information obtainer that obtains aperture pattern information corresponding to the pattern of the at least one aperture, the noise adder that adds, to the captured image, noise determined according to the aperture pattern information, and a transmitter that outputs the noise added captured image. ## Claims ### Claim 1 An image processing device comprising: an image obtainer that obtains a first captured image from a first image capturing device including a mask including at least one aperture; an information obtainer that obtains aperture pattern information corresponding to a pattern of the at least one aperture; a noise adder that adds, to the first captured image, noise determined according to the aperture pattern information; and an outputting unit that outputs the first captured image to which the noise is added. ### Claim 2 (depends on 1) The image processing device according to claim 1, wherein the noise adder adds the noise in a frequency band wider than a predetermined frequency band to the first captured image. ### Claim 3 (depends on 1) The image processing device according to claim 1, further comprising: a noise information determination unit that determines, according to the aperture pattern information, the noise to be added by the noise adder. ### Claim 4 (depends on 3) The image processing device according to claim 3, wherein when frequency characteristics of the mask based on the pattern of the at least one aperture corresponding to the aperture pattern information represent a wide frequency band including a first band and a second band higher than the first band, the noise information determination unit selects, as the noise to be added by the noise adder, at least one of salt-and-pepper noise or noise that changes an output value of a portion of the first captured image to a fixed value, and when an intensity of a frequency component of the first band is higher than an intensity of a frequency component of the second band, the noise information determination unit selects Laplace noise as the noise to be added by the noise adder. ### Claim 5 (depends on 1) The image processing device according to claim 1, wherein the mask is switchable between a first aperture pattern and a second aperture pattern different from the first aperture pattern, and the image processing device further comprises a switcher that switches the pattern of the at least one aperture of the mask between the first aperture pattern and the second aperture pattern. ### Claim 6 (depends on 1) The image processing device according to claim 1, wherein the noise includes at least one of salt-and-pepper noise, Laplace noise, white noise, pink noise, or noise that changes an output value of a portion of the first captured image to a fixed value. ### Claim 7 (depends on 1) The image processing device according to claim 1, wherein the aperture pattern information includes at least one of point spread function (PSF), a size and a shape of the at least one aperture, or information on a plurality of apertures of the mask which are the at least one aperture. ### Claim 8 (depends on 1) The image processing device according to claim 1, wherein the first image capturing device is one of a multi-pinhole camera, a lensless camera, and a coded aperture camera. ### Claim 9 (depends on 1) An identification device comprising: an image obtainer that obtains the first captured image to which the noise is added, from the image processing device according to claim 1; and an identifying unit that identifies an object shown in the first captured image to which the noise is added, by using a machine learning model trained using a dataset including an image generated by adding noise determined according to the aperture pattern information to a second captured image obtained from a second image capturing device including the mask. ### Claim 10 A method for training a machine learning model, the method comprising: obtaining a dataset including an image generated by adding noise to a captured image obtained from an image capturing device including a mask including at least one aperture, the noise being determined according to aperture pattern information corresponding to a pattern of the at least one aperture; and training the machine learning model by using the dataset obtained. ### Claim 11 An image processing method comprising: obtaining a captured image from an image capturing device including a mask including at least one aperture; obtaining aperture pattern information corresponding to a pattern of the at least one aperture; adding, to the captured image, noise determined according to the aperture pattern information; and outputting the captured image to which the noise is added.