# US 2010/0166296 A1 · Method and program for extracting silhouette image and method and program for constructing three dimensional model > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2010/0166296 A1 - Document ID: US20100166296A1 - Status: Published application - Kind: A1 - Application number: 12/630,497 - Filed: 2009-12-03 - Published: 2010-07-01 - Earliest claimed: 2008-12-26 - Assignees: Kddi Corporation - Inventors: Hiroshi SANKOH; Mehrdad PANAHPOUR TEHRANI; Sei NAITO - Web page: https://patentgrep.com/patent/US20100166296A1 ## CPC - G06T 7/12: Edge-based segmentation - G06T 7/136: involving thresholding - G06T 7/174: involving the use of two or more images - G06T 17/10: Constructive solid geometry [CSG] using solid primitives, e.g. cylinders, cubes - G06T 2207/20036: Morphological image processing - G06T 2207/20224: Image subtraction - G06T 2207/30196: Human being; Person - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract A present invention provides a method and a program for extracting the high accuracy silhouette by relatively simple process not using manual labor or special photography environment. A method for extracting the high accuracy silhouette comprises: extracting a number of first silhouettes from a number of object images and a number of background images by a background subtraction; constructing first visual hull from a number of the first silhouettes by a shape from silhouette method; constructing second visual hull by process to repair missed parts and/or to remove unwanted regions in first visual hull; and extracting a number of second silhouettes from the second visual hull. ## Claims ### Claim 1 A method for extracting a number of silhouettes representing the object from a number of object images which an object and a background are captured and a number of background images which only a background is captured comprising: a first extract step of extracting a number of first silhouettes from a number of said object images and a number of said background images by a background subtraction; a first construct step of constructing first visual hull from a number of said first silhouettes by a shape from silhouette method; a second construct step of constructing second visual hull by process to repair missed parts and/or to remove unwanted regions in first visual hull; and a second extract step of extracting a number of second silhouettes from said second visual hull. ### Claim 2 (depends on 1) The method for extracting a number of silhouettes according to claim 1, where in said second extract step from said first construct step are repeated with assuming a number of said second silhouettes to be a number of said first silhouettes at said first construct step. ### Claim 3 (depends on 1) The method for extracting a number of silhouettes according to claim 1, where in said second construct step comprises: a sub step of acquiring a number of first slice images of said first visual hull from an x-axis, a y-axis and a z-axis direction; and a sub step of performing a filter process to a number of said first slice images and constructing second visual hull according to a result of said filter process. ### Claim 4 (depends on 3) The method for extracting a number of silhouettes according to claim 3, where in said sub step of constructing second visual hull is a step of constructing said second visual hull by performing a filter process to a number of said first slice images, by searching pixels became white (representing the object region, on the other hand black represents background) by the filter process, and by repairing three dimensional coordinates of said first visual hull corresponding to said pixels. ### Claim 5 (depends on 1) The method for extracting a number of silhouettes according to claim 1, where in said second extract step comprises: a sub step of performing a filter process to a number of said first slice images, searching pixels became white by the filter process, projective-transforming said pixels into the silhouettes, and extracting a number of third silhouettes; a sub step of acquiring a number of second slice images of said second visual hull from the y-axis direction; a sub step of extracting a number of fourth silhouettes, by searching white pixels of a number of said second slice images, and by projective-transforming said pixels into the silhouettes; and a sub step of extracting a number of second silhouettes by performing a filter process to a number of said fourth silhouettes, by making white of pixels that both of a number of said fourth silhouettes performed the filter process and a number of said third silhouettes are white, and by making black of other pixels. ### Claim 6 (depends on 1) The method for extracting a number of silhouettes according to claim 1, where in said method for extracting a number of silhouettes further comprises a step of performing a filter process to a number of said second silhouettes after said second extract step. ### Claim 7 A program for extracting a number of silhouettes representing the object from a number of object images which an object and a background are captured and a number of background images which only a background is captured comprising: a first extract means for extracting a number of first silhouettes from a number of said object images and a number of said background images by a background subtraction; a first construct means for constructing first visual hull from a number of said first silhouettes by a shape from silhouette method; a second construct means for constructing second visual hull by process to repair missed parts and/or to remove unwanted regions in first visual hull; and a second extract means for extracting a number of second silhouettes from said second visual hull. ### Claim 8 A method for constructing visual hull from a number of object images which an object and a background are captured and a number of background images which only a background is captured comprising: a first extract step of extracting a number of first silhouettes from a number of said object images and a number of said background images by a background subtraction; a first construct step of constructing first visual hull from a number of said first silhouettes by a shape from silhouette method; a second construct step of constructing second visual hull by process to repair missed parts and/or to remove unwanted regions in first visual hull; a second extract step of extracting a number of second silhouettes from said second visual hull; and a third construct step of constructing third visual hull from a number of said second silhouettes by the shape from silhouette method. ### Claim 9 A program for constructing visual hull from a number of object images which an object and a background are captured and a number of background images which only a background is captured comprising: a first extract means for extracting a number of first silhouettes from a number of said object images and a number of said background images by a background subtraction; a first construct means for constructing first visual hull from a number of said first silhouettes by a shape from silhouette method; a second construct means for constructing second visual hull by process to repair missed parts and/or to remove unwanted regions in first visual hull; a second extract means for extracting a number of second silhouettes from said second visual hull; and a third construct means for constructing third visual hull from a number of said second silhouettes by the shape from silhouette method.