FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2945199417> ?p ?o ?g. }

• W2945199417 abstract "Hand posture estimation is one of the main phases of hand gestureestimation. It refers to the process of estimating a real hand imageobtained from an acquisition device in a computer. The accuracy ofthe estimated hand model has a direct impact on feature selectionand classi cation. The current hand posture estimation techniquesare divided into three classes: generative, discriminative, and hybridmethods.In the rst, a model of the hand is generated and improved to modela hand in a computer accurately. In the second class, a database ofdi erent hand images is created and used to estimate the hand model.Finally, hybrid models use both discriminative and generative methodsmostly sequentially.This thesis rst reviews the literature of hand posture estimation usinggenerative methods and identi es the current gaps. The gaps are sensitivityto hand shapes, sensitivity to a good initial posture, di culthand posture recovery in case of loss in tracking, and lack of addressingmultiple objectives to maximise accuracy and minimise computationalcost.To ll the gaps identi ed, a new hand model is proposed combining thebest features of the current 3D hand models in the literature. Therefore,the rst contribution of this thesis is the proposal of a new 3Dhand model with simple shapes and low computational complexity torender. After the proposal of the 3D hand model, it is employed to developa hand shape optimisation technique as the second contribution.The problem is formulated as a sign-objective problem with several variables and constraints.To nd the global optimum for the single-objective problem formulated,Particle Swarm Optimisation (PSO) is improved and used, as one ofthe most well-regarded optimisation algorithms in the literature withsuccessful application in both science and industry. This thesis alsodemonstrates the e ectiveness of the improved PSO in hand posturerecovery in case of tracking loss.The last contribution of this thesis is the formulation of the hand postureestimation as a bi-objective problem for the rst time in the literature.The objectives identi ed and used are to minimise the error(maximise accuracy) and minimise the number of points in the pointcloud, thus reducing the computational cost. After formulating theproblem, Multi-Objective Particle Swarm Optimisation (MOPSO) isemployed to estimate the Pareto optimal front as the solution for thisbi-objective problem.Both PSO and MOPSO were improved since it was observed that thesealgorithms are not very e cient for estimating hand postures. Therefore,their performance was improved using an evolutionary operatorcalled Evolutionary Population Dynamics (EPD). The performance ofboth techniques was tested on test functions and then applied to theproblems mentioned in the preceding paragraphs.The case studies in this thesis are 50 hand postures extracted from vestandard datasets in the literature. All the case studies were employedto benchmark the proposed 3D hand model, hand shape optimisation,and hand posture recovery. In the multi-objective section, the samecase studies were used.The results show that rstly, the proposed hand model is able to outperformthe current hand models due to the better con guration andmore uniform point cloud that it o ers. Secondly, the proposed handshape optimisation can nd an optimal shape for di erent hand sizesand promote hand personalisation. Thirdly, the improved PSO is ableto not only nd an optimal shape for the 3D hand model but also to recover from a wrong posture or tracking loss. Finally, this thesisshows that the improved MOPSO can readily estimate the Paretooptimal front for the bi-objective problem. The thesis also considersanalysing the high-dimensional results of multi-objective optimisationusing parallel coordinates to understand the relationship between theparameters and objectives of this problem for the rst time in the literature." @default.
• W2945199417 created "2019-05-29" @default.
• W2945199417 creator A5012199979 @default.
• W2945199417 date "2018-04-01" @default.
• W2945199417 modified "2023-09-23" @default.
• W2945199417 title "Evolutionary Hand Posture Estimation for Image-based Gesture Detection Systems" @default.
• W2945199417 doi "https://doi.org/10.25904/1912/2932" @default.
• W2945199417 hasPublicationYear "2018" @default.
• W2945199417 type Work @default.
• W2945199417 sameAs 2945199417 @default.
• W2945199417 citedByCount "0" @default.
• W2945199417 crossrefType "dissertation" @default.
• W2945199417 hasAuthorship W2945199417A5012199979 @default.
• W2945199417 hasConcept C111919701 @default.
• W2945199417 hasConcept C119857082 @default.
• W2945199417 hasConcept C127413603 @default.
• W2945199417 hasConcept C153180895 @default.
• W2945199417 hasConcept C154945302 @default.
• W2945199417 hasConcept C167966045 @default.
• W2945199417 hasConcept C207347870 @default.
• W2945199417 hasConcept C21200559 @default.
• W2945199417 hasConcept C24326235 @default.
• W2945199417 hasConcept C31972630 @default.
• W2945199417 hasConcept C39890363 @default.
• W2945199417 hasConcept C41008148 @default.
• W2945199417 hasConcept C97931131 @default.
• W2945199417 hasConcept C98045186 @default.
• W2945199417 hasConceptScore W2945199417C111919701 @default.
• W2945199417 hasConceptScore W2945199417C119857082 @default.
• W2945199417 hasConceptScore W2945199417C127413603 @default.
• W2945199417 hasConceptScore W2945199417C153180895 @default.
• W2945199417 hasConceptScore W2945199417C154945302 @default.
• W2945199417 hasConceptScore W2945199417C167966045 @default.
• W2945199417 hasConceptScore W2945199417C207347870 @default.
• W2945199417 hasConceptScore W2945199417C21200559 @default.
• W2945199417 hasConceptScore W2945199417C24326235 @default.
• W2945199417 hasConceptScore W2945199417C31972630 @default.
• W2945199417 hasConceptScore W2945199417C39890363 @default.
• W2945199417 hasConceptScore W2945199417C41008148 @default.
• W2945199417 hasConceptScore W2945199417C97931131 @default.
• W2945199417 hasConceptScore W2945199417C98045186 @default.
• W2945199417 hasLocation W29451994171 @default.
• W2945199417 hasOpenAccess W2945199417 @default.
• W2945199417 hasPrimaryLocation W29451994171 @default.
• W2945199417 hasRelatedWork W1761970801 @default.
• W2945199417 hasRelatedWork W1794398627 @default.
• W2945199417 hasRelatedWork W2001900846 @default.
• W2945199417 hasRelatedWork W2040854421 @default.
• W2945199417 hasRelatedWork W2250544991 @default.
• W2945199417 hasRelatedWork W2281177307 @default.
• W2945199417 hasRelatedWork W2474920480 @default.
• W2945199417 hasRelatedWork W2767753278 @default.
• W2945199417 hasRelatedWork W2886009474 @default.
• W2945199417 hasRelatedWork W2895365144 @default.
• W2945199417 hasRelatedWork W2921315464 @default.
• W2945199417 hasRelatedWork W2951049664 @default.
• W2945199417 hasRelatedWork W2953855285 @default.
• W2945199417 hasRelatedWork W2966494142 @default.
• W2945199417 hasRelatedWork W2974514154 @default.
• W2945199417 hasRelatedWork W3046723447 @default.
• W2945199417 hasRelatedWork W3159316838 @default.
• W2945199417 hasRelatedWork W3167632217 @default.
• W2945199417 hasRelatedWork W3198617239 @default.
• W2945199417 hasRelatedWork W3213561123 @default.
• W2945199417 isParatext "false" @default.
• W2945199417 isRetracted "false" @default.
• W2945199417 magId "2945199417" @default.
• W2945199417 workType "dissertation" @default.