Buluswar, S., and Draper, B. 1994. Nonparametric classification of pixels under varying outdoor illumination. Image Understanding Workshop.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....much of the hand. A threshold on the percentage of pixels in the hand segmentation is used to determine whether or not there is something in the hand. The method we use to identify skin colored pixels is based on Buluswar s work on non parametric classification of pixels under varying illumination(Buluswar Draper 1994). A multivariate decision tree (MDT) algorithm(Draper, Brodley, Utgo# 1994) is used to separate the skin colored region of the color space from the rest. Training data was taken from hand cropped images. Figures 4.2 and 4.2 show the skin detection results on two images. Figure 1: Recognizing a ....

Buluswar, S., and Draper, B. 1994. Nonparametric classification of pixels under varying outdoor illumination. Image Understanding Workshop.

....a soda can (i.e. requesting a can to be disposed of) This is done by closely examining the hand to see if it is skin colored. 3.3. 1 Identifying skin The method we use to identify skin colored pixels is based on Buluswar s work on nonparametric classification of pixels under varying illumination[2]. A multivariate decision tree (MDT) algorithm[4] is used to separate the skincolored region of the color space from the rest. Training data was taken from hand cropped images and as much data as possible was taken from regions close to the boundary of the skin colored region. Figure 3.3.1 and ....

S. Buluswar and B. Draper. Nonparametric classification of pixels under varying outdoor illumination. Image Understanding Workshop, 1994.

....uses a three stage strategy [BHP94] First, a detection process suggests regions of interest within the image worth further consideration as possible targets. An innovation at this stage, being developed at the University of Massachusetts, is the use of color as an additional detection cue [BDHR94] The second stage, being developed by Alliant Tech Systems, extends LADAR probing techniques [BJLP92] to generate target type and target pose hypotheses. Finally, given object type and pose hypotheses, an error reduction approach will generate a best fit match between the sensor and model ....

Shashi Buluswar, Bruce A. Draper, Allen Hanson, and Edward Riseman. Non-parametric Classification of Pixels Under Varying Outdoor Illumination. In Proceedings: Image Understanding Workshop, pages 1619--1626, Los Altos, CA, November 1994. ARPA, Morgan Kaufmann.

....matching algorithm requires queuing in order to function. This queuing is provided by two upstream processes: 1) target detection and 2) target type and pose hypothesis generation. Detection uses color imagery to predict targets based upon the color characteristics of the camouflaged vehicles [8]. The detection information is then passed to a target type and pose hypothesis phase which generates a list of possible target types and orientations. This hypothesis generation algorithm uses boundary template matching in the range imagery [6] Finally, for each hypothesized target, multisensor ....

....uniquely special about the other algorithms chosen to perform queuing. That said, in order to better understand the functioning of the complete system, all three components are briefly summarized. 3. 1 Target Detection The detection algorithm was developed at the University of Massachusetts [8]. Using training imagery, it learns to discriminate between color values produced by camouflaged vehicles and values produced by background terrain. 4 Actual Range is 50 meters. However range to targets are unusually short to accommodate the short operating range of the older LADAR. Wide angle ....

Shashi Buluswar, Bruce A. Draper, Allen Hanson, and Edward Riseman. Non-parametric Classification of Pixels Under Varying Outdoor Illumination. In Proceedings: Image Understanding Workshop, pages 1619-- 1626, Los Altos, CA, November 1994. ARPA, Morgan Kaufmann.

....much of the hand. A threshold on the percentage of pixels in the hand segmentation is used to determine whether or not there is something in the hand. The method we use to identify skin colored pixels is based on Buluswar s work on non parametric classification of pixels under varying illumination[2]. A multivariate decision tree (MDT) algorithm[4] is used to separate the skin colored region of the color space from the rest. Training data was taken from hand cropped images. Figures 4.2 and 4.2 show the skin detection results on two images. Figure 1. Recognizing a hand holding a can. Figure 2. ....

S. Buluswar and B. Draper. Nonparametric classification of pixels under varying outdoor illumination. Image Understanding Workshop, 1994.

....terrain. The current work uses non parametric combination of the basic red, green and blue values of the color signal and realtime detection can be supported by encoding the decision tree as a lookup table. This subproject is described further in Section 5. 3 and elsewhere in these proceedings [ Buluswar et al. 1994 ] Past work of our team has shown that bounding contours in LADAR can be used to perform recognition. Adapting and extending this work to perform target class and pose hypothesis generation is now a major project activity. Hypothesis generation places different demands on boundary contour ....

....pixel in the image as belonging to one of two classes target or non target. This determination is based upon nonparametric combinations of red, green and blue values learned by a multivariate decision tree learning procedure described below and in more detail elsewhere in these proceedings [ Buluswar et al. 1994 ] Following pixel classification, morphological techniques are used to discard single or several pixel target responses and leave target regions. Preliminary tests have shown generalization across vehicles and across lighting conditions, specifically direct sun versus overcast. To perform the ....

Shashi Buluswar, Bruce A. Draper, Allen Hanson, and Edward Riseman. Nonparametric Classification of Pixels Under Varying Outdoor Illumination. In Proceedings: Image Understanding Workshop, page (to appear), Los Altos, CA, November 1994. ARPA, Morgan Kaufmann.

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