# US 2019/0318157 A1 · Information processing apparatus, information processing method, non-transitory computer-readable storage medium > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2019/0318157 A1 - Document ID: US20190318157A1 - Status: Published application - Kind: A1 - Application number: 16/379,882 - Filed: 2019-04-10 - Published: 2019-10-17 - Earliest claimed: 2018-04-13 - Assignees: Canon Kabushiki Kaisha - Applicants: Canon Kabushiki Kaisha - Inventors: Atsuo Nomoto; Koji Ueno - Web page: https://patentgrep.com/patent/US20190318157A1 ## CPC - G06V 40/161: {Detection; Localisation; Normalisation} - G06F 18/211: Selection of the most significant subset of features - G06T 7/33: using feature-based methods - G06V 10/82: using neural networks - G06V 40/171: {Local features and components; Facial parts (eye characteristics G06V40/18); Occluding parts, e.g. glasses; Geometrical relationships} - G06V 40/172: {Classification, e.g. identification} - G06T 2200/04: involving 3D image data - G06T 2207/30201: Face - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract An information processing apparatus comprises a conversion unit configured to convert an element array, in each region set so as to partially overlap at least one set of adjacent regions in input data, into a lower dimension element array of which dimension is lower than that of the element array; a generation unit configured to generate a connected element by connecting some or all of the lower dimension element arrays converted by the conversion unit so that an overlapping portion in each of the lower dimension element arrays will be shared; and a calculation unit configured to obtain a feature amount of the input data based on convolution of the connected elements and a weight coefficient. ## Claims ### Claim 1 An information processing apparatus comprising: a conversion unit configured to convert an element array, in each region set so as to partially overlap at least one set of adjacent regions in input data, into a lower dimension element array of which dimension is lower than that of the element array; a generation unit configured to generate a connected element by connecting some or all of the lower dimension element arrays converted by the conversion unit so that an overlapping portion in each of the lower dimension element arrays will be shared; and a calculation unit configured to obtain a feature amount of the input data based on convolution of the connected elements and a weight coefficient. ### Claim 2 (depends on 1) The apparatus according to claim 1, wherein the input data is a two-dimensional image and the lower dimension element array is a one-dimensional pixel array. ### Claim 3 (depends on 2) The apparatus according to claim 2, wherein the conversion unit converts, into a one-dimensional pixel array, a two-dimensional pixel array in each region two-dimensionally set on the two-dimensional image so that adjacent regions will partially overlap each other. ### Claim 4 (depends on 3) The apparatus according to claim 3, wherein the generation unit acquires, from a one-dimensional pixel array of each succeeding region succeeding a region on one end of a region of interest column arranged in a first direction in the two-dimensional image, a portion which does not overlap a one-dimensional pixel array of a region adjacent to the succeeding region on the side of the region on one end of the region of interest column, and the generation unit generates a connected one-dimensional pixel array by connecting the portion and a one-dimensional pixel array corresponding to the region on one end. ### Claim 5 (depends on 4) The apparatus according to claim 4, wherein the calculation unit obtains the feature amount of the two-dimensional image based on the convolution of the connected one-dimensional pixel array and the weight coefficient. ### Claim 6 (depends on 2) The apparatus according to claim 2, wherein the two-dimensional image is each channel image in a multi-channel image. ### Claim 7 (depends on 6) The apparatus according to claim 6, wherein the generation unit generates a connected one-dimensional pixel array by connecting an element array obtained by arranging, in a channel image order, elements at identical positions in the one-dimensional pixel arrays corresponding to the regions on one end in the respective channel images and an element array obtained by arranging, in the channel image order, elements at identical positions in the portions corresponding to the respective channel images. ### Claim 8 (depends on 2) The apparatus according to claim 2, wherein the calculation unit performs first pooling processing on the result of the convolution of the connected one-dimensional pixel array and the weight coefficient. ### Claim 9 (depends on 8) The apparatus according to claim 8, wherein the calculation unit performs second convolution and second pooling processing on a result of the first pooling processing. ### Claim 10 (depends on 9) The apparatus according to claim 9, wherein the calculation unit obtains the feature amount by performing full connection processing on a result of the second pooling processing. ### Claim 11 (depends on 2) The apparatus according to claim 2, wherein the two-dimensional image is an image of a human face. ### Claim 12 (depends on 1) The apparatus according to claim 1, wherein the input data is a three-dimensional object obtained by stacking a plurality of two-dimensional images. ### Claim 13 (depends on 12) The apparatus according to claim 12, wherein the conversion unit converts a three-dimensional pixel array in each region set in the three-dimensional object so that adjacent regions will partially overlap each other into a two-dimensional pixel array. ### Claim 14 (depends on 13) The apparatus according to claim 13, wherein the generation unit acquires, from a two-dimensional pixel array of a succeeding region succeeding a region on one end of a region of interest column arranged in a first direction perpendicular to a stacking direction of the three-dimensional object, a portion which does not overlap a two-dimensional pixel array of a region adjacent to the succeeding region on the side of the region on one end of the region of interest column, and the generation unit generates a connected two-dimensional pixel array by connecting the portion and a two-dimensional pixel array corresponding to the region on the region on one end. ### Claim 15 (depends on 14) The apparatus according to claim 14, wherein the calculation unit obtains the feature amount of the three-dimensional object based on the convolution of the connected two-dimensional pixel array and the weight coefficient. ### Claim 16 (depends on 1) The apparatus according to claim 1, further comprising: a registration unit configured to register the feature amount and information related to an object input by a user in a registration dictionary. ### Claim 17 (depends on 16) The apparatus according to claim 16, further comprising: an authentication unit configured to authenticate the input data based on the similarity between the feature amount obtained by the calculation unit and the feature amount registered in the registration dictionary. ### Claim 18 (depends on 17) The apparatus according to claim 17, further comprising: an output unit configured to output a result of the authentication by the authentication unit. ### Claim 19 An information processing method performed by an information processing apparatus, the method comprising: converting an element array, in each region set so as to partially overlap at least one set of adjacent regions in input data, into a lower dimension element array of which dimension is lower than that of the element array; generating a connected element by connecting some or all of the lower dimension element arrays converted in the converting so that an overlapping portion in each of the lower dimension element arrays will be shared; and obtaining a feature amount of the input data based on convolution of the connected elements and a weight coefficient. ### Claim 20 A non-transitory computer-readable storage medium storing a computer program for causing a computer to function as a conversion unit configured to convert an element array, in each region set so as to partially overlap at least one set of adjacent regions in input data, into a lower dimension element array of which dimension is lower than that of the element array of a lower dimension; a generation unit configured to generate a connected element by connecting some or all of the lower dimension element arrays converted by the conversion unit so that an overlapping portion in each of the lower dimension element arrays will be shared; and a calculation unit configured to obtain a feature amount of the input data based on convolution of the connected elements and a weight coefficient.