Content uploaded by Aaron Mavrinac
Author content
All content in this area was uploaded by Aaron Mavrinac on Sep 20, 2017
Content may be subject to copyright.
A preview of the PDF is not available
A Fuzzy Associative Approach for Recognition of 3D Objects in Arbitrary Pose
Abstract and Figures
Once the human vision system has seen a 3D object from a few different viewpoints, depending on the nature of the object, it can generally recognize that object from new arbitrary viewpoints. This useful interpolative skill relies on the highly complex pattern matching systems in the human brain, but the general idea can be applied to a computer vision recognition system using comparatively simple machine learning techniques. An approach to the recognition of 3D objects in arbitrary pose relative the the vision equipment given only a limited training set of views is presented. This approach involves computing a disparity map using stereo cameras, extracting a set of features from the disparity map, and classifying it via a fuzzy associative map to a trained object.
Figures - uploaded by Aaron Mavrinac
Author content
All figure content in this area was uploaded by Aaron Mavrinac
Content may be subject to copyright.
... Several moment-based descriptors were proposed along the years, and some of them are present in this review: Ramalingam [238] presents a fuzzy surface classification paradigm, which is an extension of conventional techniques based on sign of mean and Gaussian curvatures. In his work, a fuzzy moment-based recognition technique described and tested in [275] was employed; Ong [219] presents a theoretical framework for deriving scale and translation invariants for 3-D Legendre moments through the use of direct and indirect methods, employing the obtained invariants on 3D object recognition; Xu [323] proposes a 3D object recognition method, which uses some features, color moments, texture features, Hu's moment invariants and the affine moment invariants, extracted from each 2D image of 3D objects; Mavrinac [191] presents an approach for recognition of 3D objects in arbitrary poses, providing only a limited set of training view samples. This approach involves computing a disparity map and extract, from the map, a set of disparity map features (compactness, first Hu moments and the image general distribution intensity histogram); the method presented by Wan [300] shows a classification method, based on fuzzy KNN and Bayesian Rules, to determine whether a 3D object belongs to the human class, Fig. 9 Representation of analyzed works that employ global features. ...
... Some works employed classifiers based on deep learning models [274,314,330], others used the k-nearest neighbor associated with Euclidean distance [122,158,309]. There are also a few works that utilized fuzzy [191,269], fuzzy associated with Bayesian networks, fuzzy associated with neural networks [221] and probabilistic models [237]. ...
In this paper, we present a systematic literature review concerning 3D object recognition and classification. We cover articles published between 2006 and 2016 available in three scientific databases (ScienceDirect, IEEE Xplore and ACM), using the methodology for systematic review proposed by Kitchenham. Based on this methodology, we used tags and exclusion criteria to select papers about the topic under study. After the works selection, we applied a categorization process aiming to group similar object representation types, analyzing the steps applied for object recognition, the tests and evaluation performed and the databases used. Lastly, we compressed all the obtained information in a general overview and presented future prospects for the area.
Link for the publication:
https://link.springer.com/epdf/10.1007/s10044-019-00804-4?author_access_token=paE7wTbqwKN7oCwVliHwLve4RwlQNchNByi7wbcMAY7uL2tJzq0UXA0O13kX7wvxz98EQWbRDi2uT7G5KxVe0WzCAoagCbJhmkFlrCPdZIPfyyYkaSt_0zAEiJJc2cojH9AajAmYQ5BT1LV4EonJMg%3D%3D
1 It is desirable for automated object recognition using computer vision systems to emulate the human capacity for recognition of shapes invariant to vari-ous transformations. We present an algorithm, based on a Fuzzy Associative Database approach, which uses appropriately invariant metrics and a neu-ro-fuzzy inference method to accurately classify both two-and three-dimensional objects (using different metrics for each). The system is trained using a small number of images of each object class under varying degrees of the transformations, and as we show expe-rimentally, is then able to identify objects under oth-er non-explicitly-trained degrees of the transforma-tions.
This paper is about so-called neuro-fuzzy systems, which combine methods from neural network theory with fuzzy systems. Such combinations have been considered for several years already. However, the term neuro-fuzzy still lacks proper definition, and still has the flavour of a buzzword to it. In this paper we try to give it a meaning in the context of three applications of fuzzy systems, which are fuzzy control, fuzzy classification, and fuzzy function approximation.
Surprisingly few neuro-fuzzy approaches do actually employ neural networks, even though they are very often depicted in form of some kind of neural network structure. However, all approaches display some kind of learning capability, as it is known from neural networks. This means, they use algorithms which enable them to determine their parameters from training data in an iterative process. From our point of view neuro-fuzzy means using heuristic learning strategies derived from the domain of neural network theory to support the development of a fuzzy system.
The author describes a two-stage technique for 3-D camera calibration using off-the-shelf TV cameras and lenses. The technique is aimed at efficient computation of the external position and orientation of the camera relative to the object reference coordinate system as well as its effective focal length, radial lens distortion and image scanning parameters. A critical review of the state of the art is given and it is shown that the two-stage technique has advantages in terms of accuracy, speed and versatility. A theoretical framework is established and supported by comprehensive proof Test results using real data are described. Both accuracy and speed are reported. A 388 multiplied by 480 CCD camera calibrated by this technique performed several 3-D measurement with an average accuracy of 1/4000 over the field of view, or 1/8000 over the depth. The experimental results are analyzed and compared with theoretical prediction.