# US 10,121,095 B2 · Method and device for recognizing subject area of image > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 10,121,095 B2 - Document ID: US10121095B2 - Status: Granted - Kind: B2 - Application number: 15/322,466 - Filed: 2015-06-23 - Published: 2017-10-12 - Granted: 2018-11-06 - Earliest claimed: 2014-12-30 - Assignees: Baidu Online Network Technology (Beijing) Co., Ltd. - Applicants: Baidu Online Network Technology (Beijing) Co., Ltd. - Inventors: Shijia Chen; Xubin Li; Shilei Wen; Yinan Yu - Web page: https://patentgrep.com/patent/US10121095B2 ## CPC - G06V 10/25: Determination of region of interest [ROI] or a volume of interest [VOI] - G06F 16/51: Indexing; Data structures therefor; Storage structures - G06F 16/5838: {using colour} - G06F 18/24: Classification techniques - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract A method and a device for recognizing a subject area of an image are provided. The method includes receiving an image to be recognized, and extracting a candidate subject area from the image; eliminating a non-subject area from the extracted candidate subject areas to obtain a target candidate subject area, and adjusting and classifying the target candidate subject areas to obtain at least one set of the target candidate subject areas; and merging the target candidate subject areas in each set of the target candidate subject areas to recognize the subject area of the image. ## Claims ### Claim 1 A method for recognizing a subject area of an image, comprising: receiving an image to be recognized, and extracting candidate subject areas from the image; eliminating a non-subject area from the extracted candidate subject areas to obtain target candidate subject areas, and adjusting and classifying the target candidate subject areas to obtain at least one set of the target candidate subject areas; and merging the target candidate subject areas in each set of the target candidate subject areas to recognize the subject area of the image, wherein adjusting the target candidate subject areas comprises: for each target candidate subject area, performing at least one of zooming and panning on a current target candidate subject area according to relations of sizes and of locations between a subject and a current target candidate subject area. ### Claim 2 (depends on 1) The method according to claim 1, wherein eliminating a non-subject-area from the extracted candidate subject areas comprises: recognizing the extracted candidate subject areas via classifiers with different levels to eliminate the non-subject area. ### Claim 3 (depends on 2) The method according to claim 2, wherein classifying the target candidate subject areas comprises: obtaining a grade and classification of each unadjusted target candidate subject area and each adjusted target candidate subject area respectively, and sorting the corresponding target candidate subject areas according to the grades of the target candidate subject areas in a same classification, and obtaining the set of target candidate subject areas according to a sorting result. ### Claim 4 (depends on 2) The method according to claim 2, wherein merging the target candidate subject areas in each set of the target candidate subject areas to recognize the subject area of the image comprises: merging the target candidate subject areas according to the relations of sizes and of locations between the target candidate subject areas. ### Claim 5 (depends on 1) The method according to claim 1, wherein classifying the target candidate subject areas comprises: obtaining a grade and classification of each unadjusted target candidate subject area and each adjusted target candidate subject area respectively, and sorting the corresponding target candidate subject areas according to the grades of the target candidate subject areas in a same classification, and obtaining the set of target candidate subject areas according to a sorting result. ### Claim 6 (depends on 1) The method according to claim 1, wherein merging the target candidate subject areas in each set of the target candidate subject areas to recognize the subject area of the image comprises: merging the target candidate subject areas according to the relations of sizes and of locations between the target candidate subject areas. ### Claim 7 A non-transitory storage medium for storing an application program, wherein the application program is configured to execute a method for recognizing a subject area of an image, the method comprising: receiving an image to be recognized, and extracting candidate subject areas from the image; eliminating a non-subject area from the extracted candidate subject areas to obtain target candidate subject areas, and adjusting and classifying the target candidate subject areas to obtain at least one set of the target candidate subject areas; and merging the target candidate subject areas in each set of the target candidate subject areas to recognize the subject area of the image, wherein adjusting the target candidate subject areas comprises: for each target candidate subject area, performing at least one of zooming and panning on a current target candidate subject area according to relations of sizes and of locations between a subject and a current target candidate subject area. ### Claim 8 A device for recognizing a subject area of an image, comprising: one or more processors; a memory; one or more modules, wherein the one or more modules are stored in the memory and configured to perform following operations if the one or more modules are executed by the one or more processors: receiving an image to be recognized, and extracting candidate subject areas from the image; eliminating a non-subject area from the extracted candidate subject areas to obtain target candidate subject areas, and adjusting and classifying the target candidate subject areas to obtain at least one set of the target candidate subject areas; and merging the target candidate subject areas in each set of target candidate subject areas to recognize the subject area of the image, wherein adjusting the target candidate subject areas comprises: for each target candidate subject area, performing at least one of zooming and panning on a current target candidate subject area according to relations of sizes and of locations between a subject and a current target candidate subject area. ### Claim 9 (depends on 8) The device according to claim 8, wherein eliminating a non-subject-area from the extracted candidate subject area comprises: recognizing the extracted candidate subject areas via classifiers with different levels to eliminate the non-subject area. ### Claim 10 (depends on 8) The device according to claim 8, wherein classifying the target candidate subject areas comprises: obtaining a grade and classification of each unadjusted target candidate subject area and each adjusted target candidate subject area respectively, and sorting the corresponding target candidate subject areas according to the grades of the target candidate subject areas in a same classification, and obtaining the set of target candidate subject areas according to a sorting result. ### Claim 11 (depends on 8) The device according to claim 8, wherein merging the target candidate subject areas in each set of the target candidate subject areas to recognize the subject area of the image comprises: merging the target candidate subject areas according to the relations of sizes and of locations between the target candidate subject areas.