# US 2019/0377971 A1 · Similar picture identification method, device, and storage medium > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2019/0377971 A1 - Document ID: US20190377971A1 - Status: Published application - Kind: A1 - Application number: 16/548,301 - Filed: 2019-08-22 - Published: 2019-12-12 - Earliest claimed: 2017-06-13 - Assignees: Tencent Technology (Shenzhen) Company Limited - Applicants: Tencent Technology (Shenzhen) Company Limited - Inventors: Xiangyu Song; Wei He; Guizhou Huang; Dean Guo; Qiquan Jiang - Web page: https://patentgrep.com/patent/US20190377971A1 ## CPC - G06V 10/26: Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion - G06F 18/22: Matching criteria, e.g. proximity measures - G06V 10/44: Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components - G06V 10/761: {Proximity, similarity or dissimilarity measures} - G06V 10/759: {Region-based matching} - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract A similar picture identification method is described. According to the method, processing circuitry of a device obtains n first local regions of a first picture, the n first local regions being a first set of regions that is affine invariant, and obtains m second local regions of a second picture, the m second local regions being a second set of regions that is affine invariant. The processing circuitry obtains n first characteristic values respectively corresponding to the n first local regions, and obtains m second characteristic values respectively corresponding to the m second local regions. The processing circuitry further determines, according to a comparison result of the n first characteristic values and the m second characteristic values, whether the first picture is similar to the second picture, where n and m are positive integers. ## Claims ### Claim 1 A similar picture identification method, comprising: identifying, by processing circuitry of a device, n first local regions of a first picture, the n first local regions being a first set of regions that is affine invariant; identifying, by the processing circuitry, m second local regions of a second picture, the m second local regions being a second set of regions that is affine invariant; obtaining, by the processing circuitry, n first characteristic values respectively corresponding to the n first local regions, and obtaining, by the processing circuitry, m second characteristic values respectively corresponding to the m second local regions, the first characteristic values representing feature information of the first local regions, and the second characteristic values representing feature information of the second local regions; and determining, according to a comparison result of the n first characteristic values and the m second characteristic values, whether the first picture is similar to the second picture, wherein n and m are positive integers. ### Claim 2 (depends on 1) The method according to claim 1, wherein the identifying the n first local regions of the first picture comprises: determining w first candidate regions in the first picture according to a maximally stable extremal region (MSER) algorithm, w being a positive integer; and identifying, from the w first candidate regions, the n first local regions that meet a preset condition, the preset condition comprising region sizes of the first local regions being between a first threshold and a second threshold, and the first threshold being less than the second threshold. ### Claim 3 (depends on 1) The method according to claim 1, wherein the identifying the m second local regions of the second picture comprises: determining x second candidate regions in the second picture according to a maximally stable extremal region (MSER) algorithm, x being a positive integer; and identifying, from the x second candidate regions, the m second local regions that meet a preset condition, the preset condition comprising region sizes of the second local regions being between a first threshold and a second threshold, and the first threshold being less than the second threshold. ### Claim 4 (depends on 1) The method according to claim 1, wherein the determining comprises: determining k hit regions in the first local regions, including determining an ith first local region of the n first local regions as a hit region when at least one of the second characteristic values matches a first characteristic value of the ith first local region, i being a positive integer less than or equal to n; determining a similarity score representing similarity between the first picture and the second picture according to the k hit regions, k being an integer; and determining that the first picture is similar to the second picture when the similarity score is greater than a similarity threshold. ### Claim 5 (depends on 4) The method according to claim 4, wherein the n first characteristic values are n first perceptual hash values, the m second characteristic values are m second perceptual hash values, and the determining the ith first local region as the hit region comprises: separately calculating, for a first perceptual hash value of the ith first local region, Hamming distances between the first perceptual hash value of the ith first local region and the m second perceptual hash values; and determining the ith first local region as the hit region when at least one of the calculated Hamming distances is less than a distance threshold. ### Claim 6 (depends on 4) The method according to claim 4, wherein the determining the similarity score representing the similarity between the first picture and the second picture according to k comprises: determining a ratio of k to n as the similarity score. ### Claim 7 (depends on 4) The method according to claim 4, wherein the determining the similarity score comprises: determining a ratio of an area of the k hit regions to an area of the first picture as the similarity score. ### Claim 8 A computer device, comprising: processing circuitry configured to: identify n first local regions of a first picture, the n first local regions being a first set of regions that is affine invariant; identify m second local regions of a second picture, the m second local regions being a second set of regions that is affine invariant; obtain n first characteristic values respectively corresponding to the n first local regions, and obtaining m second characteristic values respectively corresponding to the m second local regions, the first characteristic values representing feature information of the first local regions, and the second characteristic values representing feature information of the second local regions; and determine, according to a comparison result of the n first characteristic values and the m second characteristic values, whether the first picture is similar to the second picture, wherein n and m are positive integers. ### Claim 9 (depends on 8) The computer device according to claim 8, wherein the processing circuitry is configured to: determine w first candidate regions in the first picture according to a maximally stable extremal region (MSER) algorithm, w being a positive integer; and identify, from the w first candidate regions, the n first local regions that meet a preset condition, the preset condition comprising region sizes of the first local regions being between a first threshold and a second threshold, and the first preset threshold being less than the second threshold. ### Claim 10 (depends on 8) The computer device according to claim 8, wherein the processing circuitry is configured to: determine x second candidate regions in the second picture according to a maximally stable extremal region (MSER) algorithm, x being a positive integer; and identify, from the x second candidate regions, the m second local regions that meet a preset condition, the preset condition comprising region sizes of the second local regions being between a first threshold and a second threshold, and the first threshold being less than the second threshold. ### Claim 11 (depends on 8) The computer device according to claim 8, wherein the processing circuitry is configured to: determine k hit regions in the first local regions, including determining an ith first local region of the n first local regions as a hit region when at least one of the second characteristic values matches the first characteristic value of the ith first local region, i being a positive integer less than or equal to n; determine a similarity score representing similarity between the first picture and the second picture according to the k hit regions, k being an integer; and determine that the first picture is similar to the second picture when the similarity score is greater than a similarity threshold. ### Claim 12 (depends on 11) The computer device according to claim 11, wherein the n first characteristic values are n first perceptual hash values, the m second characteristic values are m second perceptual hash values, and the processing circuitry is configured to: separately calculate, for a first perceptual hash value of the ith first local region, Hamming distances between the first perceptual hash value of the ith first local region and the m second perceptual hash values; and determine the ith first local region as the hit region when at least one of the calculated Hamming distances is less than a distance threshold. ### Claim 13 (depends on 11) The computer device according to claim 11, wherein the processing circuitry is configured to: determine a ratio of k to n as the similarity score. ### Claim 14 (depends on 11) The computer device according to claim 11, wherein the processing circuitry is configured to: determine a ratio of an area of the k hit regions to an area of the first picture as the similarity score. ### Claim 15 A method of determining picture key characteristics, comprising: identifying, by processing circuitry of a device, key regions in one or more pictures; calculating, by the processing circuitry, fingerprints respectively corresponding to the picture key characteristics in the key regions; grouping the fingerprints and the key regions in association with the one or more pictures, respectively; and outputting a table that stores a result of the grouping. ### Claim 16 (depends on 15) The method according to claim 15, wherein the identifying the key regions comprises: determining w candidate regions in a picture according to a maximally stable extremal region (MSER) algorithm, w being a positive integer; and identifying, from the w first candidate regions, n local regions that meet a preset condition as a subset of the key regions, n being a positive integer, the preset condition comprising region sizes of the n local regions being between a first threshold and a second threshold, and the first threshold being less than the second threshold. ### Claim 17 (depends on 16) The method according to claim 16, wherein a region size of the n local regions ranges from 32*32 to 200*200. ### Claim 18 (depends on 16) The method according to claim 16, further comprising: converting the n local regions into n grayscale pictures of a predetermined size; performing discrete cosine transform (DCT) on the n grayscale pictures, to obtain n sets of DCT coefficient values of the n local regions; obtaining n sample regions that respectively correspond to the n local regions from the n sets of DCT coefficient values; and calculating n fingerprints corresponding to the n local regions according to the n sample regions. ### Claim 19 (depends on 18) The method according to claim 18, wherein the n fingerprints are n perceptual hash values calculated according to the n sample regions. ### Claim 20 (depends on 15) The method according to claim 15, further comprising: calculating the fingerprints according to a perceptual hash algorithm. ### Claim 21 A method of determining picture similarity, comprising: deriving a first picture characteristic grouping of a first picture by identifying first key regions in the first picture according to a maximally stable extremal region (MSER) algorithm and calculating first fingerprints respectively corresponding to the first key regions according to a perceptual hash algorithm; deriving a second picture characteristic grouping of a second picture by identifying second key regions in the second picture according to the MSER algorithm and calculating second fingerprints respectively corresponding to the second key regions according to the perceptual hash algorithm; calculating a hamming distance between the first picture characteristic grouping and the second picture characteristic grouping; and determining the picture similarity according to the calculated hamming distance.