# US 8,543,525 B2 · Method and system for the automatic detection of events in sport fields > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 8,543,525 B2 - Document ID: US8543525B2 - Status: Granted - Kind: B2 - Application number: 12/280,152 - Filed: 2007-02-28 - Published: 2009-03-19 - Granted: 2013-09-24 - Earliest claimed: 2006-03-03 - Assignees: Consiglio Nazionale Delle Richerche (CNR) - Applicants: Arcangelo Distante; Ettore Stella; Massimiliano Nitti; Liborio Capozzo; Tiziana Rita D'Orazio; Massimo Ianigro; Nicola Mosca; Marco Leo; Paolo Spagnolo; Pier Luigi Mazzeo - Inventors: Arcangelo Distante; Ettore Stella; Massimiliano Nitti; Liborio Capozzo; Tiziana Rita D'Orazio; Massimo Ianigro; Nicola Mosca; Marco Leo; Paolo Spagnolo; Pier Luigi Mazzeo - Web page: https://patentgrep.com/patent/US8543525B2 ## CPC - A63B 71/0605: {Decision makers and devices using detection means facilitating arbitration} - G06T 7/246: using feature-based methods, e.g. the tracking of corners or segments - G06V 10/147: Details of sensors, e.g. sensor lenses (fingerprint or palmprint sensors G06V40/13; vascular sensors G06V40/145; eye sensors G06V40/19) - G06V 10/24: Aligning, centring, orientation detection or correction of the image - G06V 20/42: {of sport video content} - A63B 2220/806: Video cameras - A63B 2243/0025: Football - G06T 2207/30221: Sports video; Sports image - A63B: APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT (apparatus for passive exercising, massage A61H) - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING ## Abstract The present invention refers to the problem of the automatic detection of events in sport field, in particular Goal/NoGoal events by signalling to the mach management, which can autonomously take the final decision upon the event. The system is not invasive for the field structures, neither it requires to interrupt the game or to modify the rules thereof, but it only aims at detecting objectively the event occurrence and at providing support in the referees' decisions by means of specific signalling of the detected events. ## Claims ### Claim 1 A method for evaluating an event occurrence in an area, said occurrence depending upon a mutual spatial position of at least one object with respect to a spatial position of one or more fixed references, the method comprising: a) acquiring a plurality of digital images in sequence in said area; b) for each acquired digital image of said plurality of digital images: defining a search region inside said acquired digital image; recognizing all moving objects in said acquired digital image; extracting from said acquired digital image a filtered image containing exclusively the moving objects; detecting in said search region of said filtered image the at least one object, the detecting comprising using an operator that includes a definition of a minimum and a maximum dimension associated with the at least one object; calculating the mutual spatial position based on the detecting of the at least one object; evaluating the event occurrence on the basis of the calculation; and providing an event occurence result on the basis of the evaluation, wherein said detecting is confirmed by a classifier based upon a neural network, said neural network being a) trained through wavelet coefficients and b) chosen from a library of neural networks. ### Claim 2 (depends on 1) The method according to claim 1, wherein said extracting a filtered image is carried out based upon a reference image. ### Claim 3 (depends on 2) The method according to claim 2, wherein said extracting a filtered image comprises comparing a current image to the reference image. ### Claim 4 (depends on 2) The method according to claim 2, wherein said reference image is updated over time. ### Claim 5 (depends on 1) The method according to claim 1, wherein said selecting comprises pre-processing each said acquired digital image. ### Claim 6 (depends on 5) The method according to claim 5, wherein said pre-processing comprises extracting edges of the moving objects in the image, by obtaining a bordered image. ### Claim 7 (depends on 6) The method according to claim 6, wherein said extracting the edges of the moving objects comprises constructing a correlation map by a convolution operation of the bordered image with a filter configured to characterize spatially said moving objects. ### Claim 8 (depends on 7) The method according to claim 7, wherein said filter comprises a bidimensional convolution mask. ### Claim 9 (depends on 1) The method according to claim 1, wherein said detecting further comprises calculating characteristic parameters of said moving objects. ### Claim 10 (depends on 9) The method according to claim 9, wherein said calculating is carried out by analysis of grey levels. ### Claim 11 (depends on 9) The method according to claim 9, wherein said calculating is carried out by a wavelet transformation. ### Claim 12 (depends on 1) The method according to claim 1, wherein said neural network is trained by spatial information. ### Claim 13 (depends on 12) The method according to claim 12, wherein said spatial information is obtained based upon analysis of grey levels of each acquired digital image. ### Claim 14 (depends on 1) The method according to claim 1, wherein said classifier is constructed through training steps with examples. ### Claim 15 (depends on 1) The method according to claim 1, wherein said calculating is based upon a predictive procedure. ### Claim 16 (depends on 1) The method according to claim 1, wherein said calculating the spatial position of the at least one object comprises measuring a three-dimensional position of a center of mass. ### Claim 17 (depends on 16) The method according to claim 16, wherein said measuring comprises calculating an intersection point of lines of sight generated by two observation points, ### Claim 18 (depends on 17) The method according to claim 17, wherein the lines of sight are constructed by way of homographic techniques. ### Claim 19 (depends on 1) The method according to claim 1, wherein said evaluating the event occurrence is based upon estimation of a bidimensional position inside an image plane. ### Claim 20 (depends on 1) The method according to claim 1, wherein said evaluating the event occurrence occurs in accordance to a metric decisional system. ### Claim 21 (depends on 1) The method according to claim 1, comprising storing each said acquired image. ### Claim 22 (depends on 1) A computer program, stored in a non-transitory storage means, said computer program configured to implement the method according to claim 1 when run on a computer. ### Claim 23 (depends on 1) The method according to claim 1, wherein said minimum and maximum dimension is based on at least a portion of a contour of said at least one object. ### Claim 24 (depends on 23) The method according to claim 23, wherein a) said at least one object is a ball, b) said at least a portion of the contour is a circular contour of said ball, and c) said minimum and maximum dimension is defined on the basis of said circular contour. ### Claim 25 (depends on 1) The method according to claim 1, wherein: a) said at least one object is a ball, b) said one or more fixed references is at least a portion of a goal line, and c) said event occurrence result is a Goal/NoGoal result. ### Claim 26 (depends on 25) The method according to claim 25, wherein said minimum and maximum dimension is a minimum and maximum radius of said ball. ### Claim 27 (depends on 1) The method according to claim 1, wherein said operator further comprises a gradient versor and a vector that describes a kernel of a convolution containing a normalization factor. ### Claim 28 (depends on 1) The method according to claim 1, wherein said gradient versor is operative to providing the event occurrence result independent of a contrast existing between a contour of the at least one object and a background in said acquired digital image. ### Claim 29 (depends on 1) The method according to claim 1, wherein said operator is defined as u ⁡ ( x , y ) = ∫ ∫ D ⁡ ( x , y ) ⁢ ⅇ → ⁡ ( α , β ) · O → ⁡ ( α - x , β - y ) ⁢ ⅆ α ⁢ ⅆ β 2 ⁢ ⁢ π ⁡ ( R max - R min ) wherein said maximum dimension is Rmax, said minimum dimension is Rmin, D(x,y) is a domain, {right arrow over (e)} is a gradient versor, α and β are weight parameters, and {right arrow over (O)} is a vector that describes a kernel of a convolution containing a normalization factor. ### Claim 30 A method for evaluating an event occurrence in an area, said occurrence depending upon a mutual spatial position of one or more moving subjects belonging to a predefined typology with respect to a spatial position of one or more fixed references, the method comprising: a) acquiring a plurality of digital images in sequence in said area; b) for each acquired digital image of said plurality of digital images: defining, a search region inside said acquired digital image; recognizing all moving subjects in said acquired digital image; extracting from said acquired digital image a filtered image containing exclusively the moving subjects; detecting in said search region of said filtered image, the moving subjects belonging to said predefined typology, the detecting carried out by a classifier that is based upon a neural network trained through wavelet coefficients, the neural network chosen from a library of neural networks according to spatial position of the moving subjects and visibility conditions; calculating the mutual spatial position from the detected moving subjects; evaluating the event occurrence; and signaling one of a first result or a second result on the basis of the evaluating. ### Claim 31 A system for detecting and signaling an event occurrence in an area, said occurrence depending upon mutual spatial position of one or more moving subjects belonging to a predefined typology with respect to a spatial position of one or more fixed references, said system comprising: one or more acquisition units of a plurality of digital images of said area, wherein each said one or more acquisition units comprises one or more processing units configured to define the spatial position of each said one or more moving subjects with respect to one or more fixed references and, for each acquired digital image of said plurality of digital images: i) define a search region inside said acquired digital image; ii) recognize all moving subjects in said acquired digital image; and iii) extract from said acquired digital image a filtered image containing exclusively the moving subjects; and a supervision unit for evaluating said event occurrence based upon said mutual spatial position, the evaluating comprising the use of a detection function incorporating a minimum and a maximum dimension associated with at least one subject amongst the one or more moving subjects, wherein detecting through the detection function is confirmed by a classifier based upon a neural network, said neural network being a) trained through wavelet coefficients and b) chosen from a library of neural networks. ### Claim 32 (depends on 31) The system according to claim 31, wherein said acquisition unit comprises at least one camera. ### Claim 33 (depends on 32) The system according to claim 32, wherein said at least one camera is a digital camera operating at a time resolution of 200 images/sec. ### Claim 34 (depends on 32) The system according to claim 32, further comprising means for calibrating said at least one camera. ### Claim 35 (depends on 31) The system according to claim 31, wherein each said one or more acquisition units further comprises a unit for real time recording of the acquired digital images. ### Claim 36 (depends on 31) The system according to claim 31, further comprising means for connecting said one or more acquisition units to said supervision unit. ### Claim 37 (depends on 31) The system according to claim 31, wherein said supervision unit comprises man-machine interface means. ### Claim 38 (depends on 37) The system according to claim 37, wherein said interface means is configured to allow at least one of: a) analysis of said plurality of images, or b) display a result.