A. Rojer and E. Schwartz. Design considerations for a spacevariant visual sensor with complex-logarithmic geometry. In Proc. of Int. Conference on Pattern Recognition, pages 278-- 285, 1990.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....to provide both high resolution and a wide field of view [Tsotsos, 1988] Thus future robot systems may be equipped with foveas. If so they will require vergence systems for the same reasons that humans require them. Work on spatially variant visual sensors has begun [Van der Spiegel et al. 1989; Rojer and Schwartz, 1990] so we can expect to use camera systems with foveas in the near future. Motion Blur: Motion blur degrades the resolution of moving images. An image that moves over the retina will be degraded by motion blur, according to the integration time of the receptors (which have limited spatiotemporal ....

Alan S. Rojer and Eric L. Schwartz, "Design Considerations for a Space-Variant Visual Sensor with Complex-Logartihmic Geometry," In Proc. of the International Conference on Pattern Recognition, Philadelphia, PA, June 1990.

....Information from Cognitive Processing is reported as the model s final output, but it is also used to bias future fixation point choices. It has been estimated that processing the whole retinal image at foveal resolution would increase the computational cost by four orders of magnitude [4]. Most current early vision models are subsymbolic: low level ( early ) vision is primarily stimulus driven and relatively isolated from higher level ( later ) processes, which are possibly symbolic. Cognitive Processing Peripheral Feature Extractor Saccade Generator Selection Foveal ....

) Rojer, A.S. & Schwartz, E.L. "Design considerations for a space-variant visual sensor with complex-logarithmic geometry", 10 International Conference on Pattern Recognition, Vol.2, pp. 278-285, 1990.

....index, and the RF rays are indexed from Gamma=2 to =2 in the RHS, and from 3=2 to =2 in the LHS. The correspondence of the positive half of the vertical midline of (a) is indicated in (b) vertical midline. The RF values are also averages of the intensity values of all pixels within RF areas [53]. With this version of the mapping, the singularity problem, the need for a uniform resolution patch, and the fovea periphery boundary problem are all eliminated. However, neighbouring RF s on each side of the midline in the input image are no longer adjacent in the cortical image. This ....

....along the vertical midline. The right hand side (RHS) is mapped using w = log(z a) which produces the right wing of the butterfly image. The left hand side (LHS) is mapped using the complementary expression w = 2 log(a) Gamma log( Gammaz a) which produces the left wing of the output image [53]. The combined mapping is conformal within each half plane. In a strict mathematical sense, the properties of scale and rotation invariance are not present in this 5 It should be noted that the connectivity graph framework can be applied to any kind of space variant sensor including those ....

A. S. Rojer and E. L. Schwartz, "Design considerations for a space-variant visual sensor with complex logarithmic geometry," in IEEE - Proc. 10th International Conference on Pattern Recognition, vol. 2, pp. 278--285, 1990.

....(b) we show a uniform image and a foveated version of the same image. The process of going from (a) to (b) is called foveating the image (a) One of the most interesting forms of foveated images is based on the complex logarithm function. Such logmap images were studied by Rojer and Schwartz [18] and others. The complex logmap is a model consistent with empirical data on the mapping from primate retina to the visual cortex [19, 20] This neuro physiological model goes back to the pioneering (and the Nobel prize) work of Hubel and Wiesel [6, 7] Rojer [17] demonstrated many favorable ....

Alan S. Rojer and E.L. Schwartz. Design considerations for a space-variant visual sensor with a complex-logarithmic sensor geometry. In 10th ICPR, volume 2, pages 278--285, 1990.

....into consideration by the computer vision community. This is evident from recent work on active vision systems and heads [18, 11, 31] and general active vision concepts and algorithms [2, 6, 1, 7, 12] One of the fundamental features of active vision is the use of space variant vision and sensors [37, 34, 32], that allow, in the case of the log polar representation data reduction as well as a certain degree of size and rotation invariance. The use of such sensors require eOEcient mechanisms for gaze control, that are, in turn, directed by attentional algorithms. Using psychophysical terms, these ....

A. Rojer and E. Schwartz. Design considerations for a space-variant visual sensor with complex logarithmic geometry. In Proceedings of the 10th IAPR International Conference on Pattern Recognition, pages 278285, 1990.

....(b) we show a uniform image and a foveated version of the same image. The process of going from (a) to (b) is called foveating the image (a) One of the most interesting forms of foveated images is based on the complex logarithm function. Such logmap images were studied by Rojer and Schwartz [19] and others. The complex logmap is a model consistent with empirical data on the mapping from primate retina to the visual cortex [21, 22] This neuro physiological model goes back to the pioneering (and the Nobel prize) work of Hubel and Wiesel [6, 7] Rojer [18] demonstrated many favorable ....

Alan S. Rojer and E.L. Schwartz. Design considerations for a space-variant visual sensor with a complex-logarithmic sensor geometry. In 10th ICPR, volume 2, pages 278--285, 1990.

....section, a brief overview of the two classes of retinal models found in the literature is presented. The two classes are called here conformal mapping models and overlapping circular receptive fields models. Both types have the property of scale and rotation invariance in their peripheral outputs [20, 21, 22] and assume square input pixels. This overview leads to a presentation of the mapping model used in our system and the output representation selected. The two conformal retinal mapping models are based on the complex logarithm function: w = log z [23] and w = log(z a) 21] Complex variables z ....

....outputs [20, 21, 22] and assume square input pixels. This overview leads to a presentation of the mapping model used in our system and the output representation selected. The two conformal retinal mapping models are based on the complex logarithm function: w = log z [23] and w = log(z a)[21]. Complex variables z and w represent pixel coordinates in the input and output images, respectively. For the log(z) model, the input image mapping template consists of a log polar grid: a set of rays emanating from the centre of the image at regular angular increments, and a set of circles where ....

A. S. Rojer and E. L. Schwartz, "Design considerations for a space-variant visual sensor with complex logarithmic geometry," in IEEE - Proc. 10th International Conference on Pattern Recognition, vol. 2, pp. 278--285, 1990.

....vision. In figure 1 we show a foveated version of an uniform image. The process of going from an uniform image to figure 1 is called foveation . One of the most interesting forms of foveated images is based on the complex logarithm function. Such logmap images were studied by Rojer and Schwartz [10] and others. The complex logmap is a model consistent with empirical data on the mapping from primate retina to the visual cortex [11, 12] This neuro physiological model goes back to the pioneering (and the Nobel prize) work of Hubel and Wiesel [5, 6] Rojer [9] demonstrated many favorable ....

Alan S. Rojer and E.L. Schwartz. Design considerations for a space-variant visual sensor with a complexlogarithmic sensor geometry. In 10th ICPR, volume 2, pages 278--285, 1990.

....space variant sampling realizes an excellent compromise between information accuracy in the fovea and width of the field of view. This peculiarity is well described by figure 2 where traditional and space variant sampling are compared using a limited number of pixels. Moreover, recent results [Rojer and Schwartz, 1990; Deitrich, 1991; Sandini and Tistarelli, 1991] demonstrate that, due to the mathematical properties of log polar mapping, space variant sensing is often appropriate for robotic implementations and task driven motor control; for instance it is very interesting in scene analysis, object ....

A.S. Rojer and E.L. Schwartz, "Design Consideration for a Space Variant Visual Sensor with Complex Logarithmic Geometry," In Proc.Int.Conf. on Pattern Recognition, Philadelphia, 1990.

....be distorted. The solution used in the visual system of all higher mammals is to use space variant resolution. A high resolution fovea consists of a small area in the middle of the field of view surrounded by a low resolution periphery . This leads to a data compression of 1:10,000 in humans (Rojer Schwartz, 1990). The drawback is that some additional mechanism is required to rapidly move the sensor so each object can be re analyzed foveally once it has been segmented. Once again, a biomorphic solution is available: the ocular motor system evolved as a solution to the same problem in mammals. This approach ....

Rojer, A. S., & Schwartz, E. L. (1990). Design Considerations for a Space-Variant Visual Sensor with Complex-Logarithmic Geometry. 10th International Conference on Pattern Recognition, 2, 278-285.

....a wide field of view [Tsotsos, 1988] Thus future robot systems may be equipped with foveas. If so they will require pursuit and vergence systems for the same reasons that natural foveal visual systems require them. Work on spatially variant visual sensors has begun [Van der Spiegel et al. 1989; Rojer and Schwartz, 1990] so we can expect to use camera systems with foveas in the near future. Motion Blur: Motion blur degrades the resolution of moving images. An image that moves over the retina will be degraded by motion blur, according to the integration time of the receptors (which have limited spatiotemporal ....

Alan Rojer and Eric Schwartz, "Design Considerations for a Space-Variant Visual Sensor with Complex-Logartihmic Geometry," In Proc. of the International Conference on Pattern Recognition, Philadelphia, Pennsylvania, June 1990.

.... as follows: ffl Significant data reduction is achieved through a non uniform, foveated sampling scheme at the receptor level and greater convergence of cone inputs into ganglion cell outputs with increasing eccentricity [19] 13] 28] Such sampling schemes may be modelled using log polar mappings [29][48] ffl The cone transduction function is described by a log like function with saturation nonlinearities and a dynamic range of 3 log units in intensity [11] 6] Mechanisms within the cone itself and network feedback from other retinal cells locally adapt the sensitivity of cone photoreceptors ....

.... Although one group has developed a VLSI foveated sensor, the number of photosensitive elements in this sensor is rather small [30] Foveated sampling may also be mimicked by uniform sampling CCD cameras using log polar transformations on the input image before further processing is performed [29][48] 51] 4] The motivation for such mappings is based on the close fit of the log polar mappings of the visual field into V1 of the cortex [29] Using a foveated sampling scheme for data reduction not only reduces the computations required at the retinal or sensor level, but has even greater ....

[Article contains additional citation context not shown here]

A.S. Rojer and E.L. Schwartz. Design considerations for a space-variant visual sensor with complex-logarithmic geometry. In Proceedings of 10th International Conference on Pattern Recognition, volume 2, pages 278--285, 1990.

.... fixed its attention to some location in the scene and that 1 The term active sensing, as used here, should not be confused with the term active sensor (devices that transmit a signal in order to make a reading) the sensor used is employing a variable resolution imaging scheme such as [19][2] The eye movements that shift the focus of attention, called saccades, can be guided by low resolution visual cues, i.e. motion or color as described in [24] Once the agent s attention has been drawn to a part of the scene, the next step is to transform the sensory information into a robot ....

Alan S. Rojer and Eric L. Schwartz, Design considerations for a space-variant visual sensor with complex-logarithmic geometry, In Proceedings of the international conference on pattern recognition, 1990

....into consideration by the computer vision community. This is evident from recent work on active vision systems and heads [18, 11, 31] and general active vision concepts and algorithms [2, 6, 1, 7, 12] One of the fundamental features of active vision is the use of space variant vision and sensors [37, 34, 32], that allow, in the case of the log polar representation data reduction as well as a certain degree of size and rotation invariance. The use of such sensors require efficient mechanisms for gaze control, that are, in turn, directed by attentional algorithms. Using psychophysical terms, these ....

A. Rojer and E. Schwartz. Design considerations for a space-variant visual sensor with complex logarithmic geometry. In Proceedings of the 10th IAPR International Conference on Pattern Recognition, pages 278--285, 1990.

....favor of using space variant sensors on board of an active vision system is that they might improve computational capabilities. This is true not only in relation to the small amount of data coming from the sensors but also in relation to the peculiarities of the log polar mapping. Recent results [12, 5, 14] show that non uniform sampling sensors are often suitable for vision applications and task specific achievement (autonomous navigation, vergence control, time to impact estimate, object tracking and object recognition are some examples) 3 An Active Space Variant Camera Mount The fact of having ....

.... we obtain the result shown in Figure 6; as in the computation of the temporal derivatives, the moving object 2 The term cortical image derives from the fact that the log polar mapping obtained by the space variant sensor is similar to the retino cortical mapping of human visual system [12]. MATROX IM 640 Image Processing Board PC 486 EISA bus Host Computer KLINGER MM 2000 Axis Controller Board EISA bus IMAGE CLD IMAGE RTP IMAGE 640 Image BUS Host BUS Base Board Color Acquisition Real Time Processor Figure 4: The setup used in the experimental phase: in the right side a detail of ....

A.S. Rojer and E.L. Schwartz. Design considerations for a space variant visual sensor with complex logarithmic geometry. In Proc. Int. Conf. on Pattern Recognition, Philadelphia, PA, 1990.

No context found.

Rojer, A. S. and Schwartz, E. L. (1990). Design considerations for a space-variant visual sensor with complex-logarithmic geometry. 10th International Conference on Pattern Recognition, Vol. 2, pages 278--285.

No context found.

A. S. Rojer and E. L. Schwartz. Design considerations for a space-variant visual sensor with complex-logarithmic geometry. 10th International Conference on Pattern Recognition, Vol. 2, pages 278--285, 1990.

No context found.

Rojer, A. S. and Schwartz, E. L. (1990). Design considerations for a spacevariant visual sensor with complex-logarithmic geometry. 10th International Conference on Pattern Recognition, Vol. 2, pages 278--285.

No context found.

A. S. Rojer and E. L. Schwartz, "Design considerations for a space-variant visual sensor with complex-logarithmic geometry," 10th International Conference on Pattern Recognition, Vol. 2, pp. 278--285, 1990.

No context found.

A. S. Rojer and E. L. Schwartz, \Design considerations for a spacevariant visual sensor with complex-logarithmic geometry," 10th International Conference on Pattern Recognition, Vol. 2, pp. 278-285, 1990.

....the instability is manifest are excluded from the training set. Even on a serial architecture, the combination of the GFA algorithm with some type of spacevariant sensor geometry would yield an early vision system that can be run in real time. For example, using a complex logarithmic geometry (Rojer Schwartz, 1990) the diffusion algorithm outlined in this paper only requires the modification of the r operator to be weighted by a negative exponential of the radial coordinate. The remainder of the GFA algorithm is unchanged, yielding another one to two more orders of magnitude increase in execution speed. ....

Rojer, A. S., & Schwartz, E. L. (1990). Design considerations for a space-variant visual sensor with complex-logarithmic geometry. In 10th International Conference on Pattern Recognition, Vol. 2, pp. 278--285.

No context found.

A. Rojer and E. Schwartz. Design considerations for a spacevariant visual sensor with complex-logarithmic geometry. In Proc. of Int. Conference on Pattern Recognition, pages 278-- 285, 1990.

No context found.

A.S. Rojer and E.L. Schwartz. Design considerations for a space-variant visual sensor with complex-logarithmic geometry. In Proceedings of the International Conference on Pattern Recognition, pages 278--285, 1990. Referred to in page(s) 11, 12, 36, 37, 59, 60, 157.

No context found.

A. S. Rojer and E. L. Schwartz, "Design Considerations for a Space-Variant Visual Sensor with Complex-Logarithmic Geometry," In 10th International Conference on Pattern Recognition, 2, pp. 278-285, 1990.

No context found.

Rojer A, Schwartz EL (1990) Design considerations for a space-variant visual sensor with complex -logarithmic geometry, 10th International Conference on Pattern Recognition, 2, 278-285.

No context found.

A.S. Rojer and E.L. Schwartz. Design considerations for a space variant visual sensor with complex logarithmic geometry. In Proc. Int. Conf. on Pattern Recognition, Philadelphia, PA, 1990.

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