Stefan Lanser, Olaf Munkelt, and Christoph Zierl. Robust video-based object recognition using CAD models. In U. Rembold, R. Dillmann, L. O. Hertzberger, and T. Kanade, editors, Intelligent Autonomous Systems IAS-4, pages 529-- 536. IOS Press, 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....be computed without loss of accuracy. Furthermore, since no explicit region needs to be constructed and because the border of a region usually consists of much less points than the entire region, the approach is very efficient. The presented algorithm will be used in the vision system described in [3, 4, 5] to approximate edges by straight lines and ellipse segments using an approach similar to [8] 10 A Calculation of the First Order Moments The integral for ff x in (33) can be decomposed into a sum by (25) Each term is given by: Z b i Gammaxy dx = Gamma 1 Z 0 x i (t)y i (t)x 0 i (t) ....

Stefan Lanser, Olaf Munkelt, and Christoph Zierl. Robust video-based object recognition using CAD models. In U. Rembold, R. Dillmann, L.O. Hertzberger, and T. Kanade, editors, Intelligent Autonomous Systems IAS-4, pages 529--536. IOS Press, 1995.

....object identification. If none of the known object classes can be matched, the object is inserted into the model as part of the background. This at least prevents collisions and further mismatches. The applied algorithms for video based object identification and localization are described e.g. in [1, 4]. Predictions corresponding to single object classes are requested from the model and matched with the sensor data. In addition to the model update done by the various perception tasks information can also be updated on a symbolic level. Independently operating robots communicate about ....

Stefan Lanser, Olaf Munkelt, and Christoph Zierl. Robust video-based object recognition using cad models. In U. Rembold, R. Dillmann, L.O. Hertzberger, and T. Kanade, editors, Proc Conf. Intelligent Autonomous Systems, pages 529--536. IOS Press, 1995.

....thus being an ideal application for testing and illustrating the model. An experimental example is described in section 3.3. In the experimental framework, several algorithms for object identification have been examined. All have the need for feature predictions for the assumed object in common [2, 10]. The accessor is focussed on object classes and poses are supplied in class relative coordinate systems. If an object finally is identified, a new instance is created and inserted into the model. In addition to these sensor specific access channels, information can be retrieved and ....

S. Lanser, O. Munkelt, und C. Zierl. Robust videobased object recognition using cad models. In U. Rembold, R. Dillmann, L. Hertzberger, und T. Kanade, editors, Proc Conf. Intelligent Autonomous Systems, pages 529--536. IOS Press, 1995.

....without loss of accuracy. Furthermore, since no explicit region needs to be constructed, and because the border of a region usually consists of significantly fewer points than the entire region, the approach is very efficient. The presented algorithm will be used in the vision system described in [5, 6, 7] to approximate edges by straight lines and ellipse segments using an approach similar to [13] A Proof of Proposition 1 We will calculate the coefficients b p 1;q 1 k;l of x k i x p 1 Gammak i Gamma1 y l i y q 1 Gammal i Gamma1 if we expand (26) and (21) In the first case we have: b ....

Stefan Lanser, Olaf Munkelt, and Christoph Zierl. Robust video-based object recognition using CAD models. In U. Rembold, R. Dillmann, L.O. Hertzberger, and T. Kanade, editors, Intelligent Autonomous Systems IAS-4, pages 529--536. IOS Press, 1995.

....the model as part of the background. This at least prevents collisions and further mismatches. 3.3 Object Identification In the experimental framework, several algorithms for object identification have been examined. All have the need for feature predictions for the assumed object in common [3, 6]. The accessor is focussed on object classes and poses are supplied in class relative coordinate systems. If an object finally is identified, a new instance is created and inserted into the model. 3.4 State Identification The task with the most model interactions is the identification of yet ....

S. Lanser, O. Munkelt, and C. Zierl. Robust videobased object recognition using cad models. In U. Rembold, R. Dillmann, L. Hertzberger, and T. Kanade, editors, Proc Conf. Intelligent Autonomous Systems, pages 529--536. IOS Press, 1995.

....driven. This contributioncan also be seen in the context of relational matching [11, 12] 2. System Architecture This section presents a short overview about the main modules of our system MORAL 1 which is used for experiments in the field of vision based object recognition and localization [4]. Using a single intensity image, the aim of the recognition task is to construct an interpretation [2] I = h object; f(I j1 ; M i 1 ) I jk ; M i k )g; R; t) i (1) with object the object hypothesis, I j l ; M i l ) the correspondence between image feature I j l and model feature M i ....

S. Lanser, O. Munkelt, and C. Zierl. Robust Video-based Object Recognition using CAD Models. In Intelligent Autonomous Systems IAS-4, pp. 529--536. IOS Press, 1995.

....ok = moral rec object(ID, Object, Pose) a) b) Fig. 1. a) Overview of MORAL and (b) a typical RPC sequence. The object recognition system MORAL presented in this contribution accomplishes these tasks by comparing the predicted model features with the features extracted from a video image [LMZ95] The underlying 3D models are polyhedral approximations of the environment provided by a hierarchical world model. This contribution mainly deals with rigid objects. The extension of this work to the recognition of articulated objects, i.e. objects consisting of multiple rigid components ....

S. Lanser, O. Munkelt, and C. Zierl. Robust Video-based Object Recognition using CAD Models. In IAS-4, pages 529--536. IOS Press, 1995.

....as sensor combined with appropriate computer vision methods. Given a proper model of its environment, an AMS can perform self localization as well as pose estimation with a single monocular video image by aligning model features like 3D points or 3D lines with the corresponding image features [1]. A lot of work has been done to establish these correspondences [2, 3, 4] and thus to compute the world position of the camera or the relative pose of an object, respectively [5, 6, 7] The search for correspondences between model and image features has exponential complexity O(n m ) with n ....

....measurements of various sensors (CCD cameras, laser scanners, radar, ultrasonic sensors, etc. and dead reckoning. In our case another source of uncertainty information is an object recognition module determining which view out of 320 possible 2D views of an object is present in the image [1]. The correct pose might be between neighbouring views which can be modelled by an appropriate covariance matrix describing the transformation of viewpoints from one view to its neighbours. Model uncertainties are available from the 3D reconstruction process which has gathered the model ....

S. Lanser, O. Munkelt, and C. Zierl. Robust Video-based Object Recognition using CAD Models. In U. Rembold, R. Dillmann, L.O. Hertzberger, and T. Kanade, editors, Intelligent Autonomous Systems IAS-4, pages 529--536. IOS Press, 1995.

....process, and a single intensity image of the scene. The pose estimation is performed in two steps: First, hypotheses of visible objects and their rough pose are generated by a recognition module. In a second step, these hypotheses are verified and refined by a localization module. For details see [4]. In case of multiple instances of the same object appearing in a scene, this process can be iterated. After each iteration the image features already mapped to previously detected objects are eliminated. Model Generation. Using a tesselated Gaussian sphere each object is represented by a set of ....

S. Lanser, O. Munkelt, and C. Zierl. Robust Video-based Object Recognition using CAD Models. In U. Rembold, R. Dillmann, L.O. Hertzberger, and T. Kanade, editors, Intelligent Autonomous Systems IAS-4, pages 529--536. IOS Press, 1995.

....ones. The joint state of the door wing is unknown, so the corresponding mask is predicted, hiding features of the door frame. 3 Hierarchical Object Recognition This section presents our approach to vision based 3D object recognition using single images captured by a monocular CCD sensor [10]. The basic principle is to establish correspondences between detected image features and appropriate model features provided by the framework described in the previous section. The recognition of solid objects, i.e. objects without joints or with known joint configuration, can be formalized by ....

....of rigid objects is based on a set of characteristic 2D views (mul tiview representation) called CVs. The determination of the viewpoints defining the CVs is based on the triangulated Gaussian sphere which guarantees an approximately homogeneous distribution of viewpoints around the object [10]. Alternatively, an aspect based approach to select object specific viewpoints could be employed [17] The CVs are predictions of model features (sensor view) provided by the model. Knowledge about the current scene can be used to further restrict the number of CVs to be considered: For example, ....

S. Lanser, O. Munkelt, and C. Zierl. Robust Videobased Object Recognition using CAD Models. In Intelligent Autonomous Systems IAS-4, pages 529--536. IOS Press, 1995.

No context found.

Stefan Lanser, Olaf Munkelt, and Christoph Zierl. Robust video-based object recognition using CAD models. In U. Rembold, R. Dillmann, L. O. Hertzberger, and T. Kanade, editors, Intelligent Autonomous Systems IAS-4, pages 529-- 536. IOS Press, 1995.

No context found.

Stefan Lanser, Olaf Munkelt, and Christoph Zierl. Robust video-based object recognition using CAD models. In U. Rembold, R. Dillmann, L. O. Hertzberger, and T. Kanade, editors, Intelligent Autonomous Systems IAS-4, pages 529-- 536. IOS Press, 1995.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC