초록 ▼

Disclosed is a method and apparatus for searching a human three-dimensional (3D) posture which can rapidly and accurately search a human 3D posture by effectively representing the human 3D posture with a small amount of data. According to the method, a posture database is established by extracting p

Disclosed is a method and apparatus for searching a human three-dimensional (3D) posture which can rapidly and accurately search a human 3D posture by effectively representing the human 3D posture with a small amount of data. According to the method, a posture database is established by extracting posture descriptors using relation among respective joints of a body, and a feature of a query 3D posture is simultaneously extracted in the same manner. Then, the similarity is calculated by comparing the posture descriptor of the query 3D posture with the posture descriptors of postures in the posture database, and then outputted to search the human 3D posture.

대표청구항 ▼

Disclosed is a method and apparatus for searching a human three-dimensional (3D) posture which can rapidly and accurately search a human 3D posture by effectively representing the human 3D posture with a small amount of data. According to the method, a posture database is established by extracting p

Disclosed is a method and apparatus for searching a human three-dimensional (3D) posture which can rapidly and accurately search a human 3D posture by effectively representing the human 3D posture with a small amount of data. According to the method, a posture database is established by extracting posture descriptors using relation among respective joints of a body, and a feature of a query 3D posture is simultaneously extracted in the same manner. Then, the similarity is calculated by comparing the posture descriptor of the query 3D posture with the posture descriptors of postures in the posture database, and then outputted to search the human 3D posture. patents is selected from the group consisting of either: an individual selected patent, a group of commonly owned patents, a portfolio of patents controlled by one or more public corporations, a portfolio of patents controlled by one or more pre-IPO companies, all patents listing one or more particular named inventors, all patents naming one or more particular prosecuting attorneys or law firms, all patents classified within one or more PTO patent classifications, or all patents issued between a first date and a second date. 12. The method of claim 1 wherein said selected patent metrics comprise one or more characteristics of each said patents in said first and second patent populations that are determined or assumed to have either a positive or negative correlation with the presence or absence of said first or second quality to a statistically significant degree. 13. The method of claim 12 wherein said selected patent metrics include one or more of the following: number of claims per patent, number of words per claim, different words per claim, length of patent specification, number of drawing pages or figures, number of cited prior art references, age of cited references, number of subsequent citations received, subject matter classification and sub-classification, origin of the patent, payment of maintenance fees, name of prosecuting attorney or law firm, examination art group, or length of pendency in the PTO. 14. The method of claim 12 wherein at least one of said patent metrics include one or more of the following: patent marking data, claim relatedness, patent relatedness, or claim type. 15. The method of claim 12 wherein at least one of said patent metrics comprises a modified claim word-count metric whereby each word and/or word phrase in a patent claim of interest is assigned a certain point value generally proportional to its determined frequency of use in a relevant patent population and wherein the word-count metric is set equal to the sum of each of the individual word point values for essentially all of the words or word phrases contained within said claim. 16. The method of claim 12 wherein at least one of said selected patent metrics comprises a relatedness metric generally indicative of the commonality of word or word phrase usage between one or more patent claims and/or patent specifications. 17. The method of claim 1 wherein said regression model comprises a multiple regression model that correlates multiple individual predictor variables comprising said selected patent metrics to a single desired criterion variable comprising the desired output rating or ranking. 18. The method of claim 17 wherein said multiple regression model has the form: CVm=.function.{PV1,PV2. . . PVn} where: CVm=criterion variable or quality/event desired to be predicted PVn=predictor variables or selected patent metrics. 19. The method of claim 18 wherein said regression model includes no more than about 10 to 30 predictor variables. 20. The method of claim 19 wherein said regression model includes between about 15 and 25 predictor variables. 21. The method of claim 1 wherein said rating or ranking is generally predictive of the probability of the patents in the third population being found either valid or invalid, being found either infringed or not infringed, or being maintained in force beyond a predetermined time period. 22. The method of claim 1 comprising the further step of determining the statistical accuracy of the regression model in accordance with the general formula: SA(m)=CO/(CO+IN) where: SA(m)=statistical accuracy of regression model (m) CO=number of correct predictions for model (m) IN=number of incorrect predictions for model (m). 23. The method of claim 22 comprising the further steps of: incrementally modifying the regression model (m) to produce a modified regression model (m+1); determining the statistical accuracy of the modi