MUMBAI, India, June 27 -- Intellectual Property India has published a patent application (202517054021 A) filed by Google Llc, Mountain View, U.S.A., on June 4, for 'reductive iteration convolution-based keypoint detection.'
Inventor(s) include Nordrum, Josephine; and Sammoura, Firas.
The application for the patent was published on June 27, under issue no. 26/2025.
According to the abstract released by the Intellectual Property India: "A method includes obtaining an input image, and determining a convolution image by convolving the input image with a kernel configured to detect an image feature. The method also includes selecting, from the convolution image, a keypoint pixel based on a corresponding value thereof being within a detection value range. The keypoint pixel represents a location within the input image of an instance of the image feature. The method additionally includes updating the convolution image by assigning a neutral value to the keypoint pixel and neighboring pixels positioned at predetermined coordinates relative to the keypoint pixel. The neutral value is excluded from the detection value range. The plurality of neighboring pixels defines a minimum spacing between the keypoint pixel and subsequent keypoint pixels detected based on the convolution image as updated. The method further includes generating an output data that represents the keypoint pixel and the subsequent keypoint pixels."
The patent application was internationally filed on Dec. 27, 2022, under International application No.PCT/US2022/082414.
Disclaimer: Curated by HT Syndication.
