William T. Freeman and Edward H. Adelson. The Design and Use of Steerable Filters. Transactions on Pattern Analysis and Machine Intelligence, vol. 13, no. 9, pp. 891-906, September 1991.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....an object recognition algorithm using coloured receptive fields [4] The basis for the receptive field are Gaussian derivatives in order to profit from scalability and the possible orientation to any arbitrary direction. These two properties allow recognition independent from scale and orientation [6, 3]. The coloured receptive fields can be adapted to the recognition problem by the selection of different derivatives in the luminance and chrominance channels. An example of a coloured receptive field used for this application is shown in figure 1. The receptive field is oriented vertically with ....

W.T. Freeman and E.H. Adelson. The design and use of steerable filters. Transactions on Pattern Analysis and Machine Intelligence, 13(9):891--906, September 1991.

....g(x; dx n Hen x g(x; 2) with g(x; p 2 e x 2 2 2 where Hen stands for the n Hermite type e polynomials [1] Gaussian derivatives have an explicit scale parameter, and can though be generated at any scale. With steerable filters proposed by Freeman [6] Gaussian derivatives can be oriented in any arbitrary direction. With automatic scale selection [9] the local scale of a feature can be determined. The object in an image can be normalised by scale which allows recognition under scale changes. The determination of the dominant orientation of a ....

....from several points allow to estimate the pose of the object. The approach based on Gaussian derivatives proposed in [3] serves as benchmark for the evaluation of the results. This approach is fast due to efficient storage and recursive filters [14] rotation invariant due to steerable filters [6], invariant to scale due to automatic scale selection [9] and robust to occlusions due to receptive fields. It produces good results for compact textured objects (see section 5.1) The approach 4 Figure 1: An image as a surface in a subspace of R fails completely for objects with sparse texture ....

W.T. Freeman and E.H. Adelson. The design and use of steerable filters. Transactions on Pattern Analysis and Machine Intelligence, 13(9):891--906, September 1991. 11

....g(x; dx n = n Hen x g(x; 2) with g(x; p 2 e x 2 2 2 where Hen stands for the n Hermite type e polynomials [1] Gaussian derivatives have an explicit scale parameter, and can though be generated at any scale. With steerable lters proposed by Freeman [6] Gaussian derivatives can be oriented in any arbitrary direction. With automatic scale selection [9] the local scale of a feature can be determined. The object in an image can be normalised by scale which allows recognition under scale changes. The determination of the dominant orientation of a ....

....from several points allow to estimate the pose of the object. The approach based on Gaussian derivatives proposed in [3] serves as benchmark for the evaluation of the results. This approach is fast due to ecient storage and recursive lters [14] rotation invariant due to steerable lters [6], invariant to scale due to automatic scale selection [9] and robust to occlusions due to receptive elds. It produces good results for compact textured objects (see section 5.1) The approach fails completely for objects with sparse texture or objects of small sizes or with holes. The reason is ....

W.T. Freeman and E.H. Adelson. The design and use of steerable lters. Transactions on Pattern Analysis and Machine Intelligence, 13(9):891-906, September 1991.

....Hen i x oe j g(x; oe) 2) with g(x; oe) 1 p 2oe e Gamma x 2 2oe 2 where Hen stands for the n th Hermite type e polynomials [AS65] Gaussian derivatives have an explicit scale parameter, oe, and can though be generated at any scale. With steerable filters proposed by Freeman [FA91] Gaussian derivatives can be oriented in any arbitrary direction. With automatic scale selection [Lin98] the local scale of a feature can be determined. The object in an image can be normalised by scale which allows recognition under scale changes. The determination of the dominant orientation of ....

....several points allow to estimate the pose of the object. The approach based on Gaussian derivatives proposed in [Col99] serves as benchmark for the evaluation of the results. This approach is fast due to efficient storage and recursive filters [YvV95] rotation invariant due to steerable filters [FA91], invariant to scale due to automatic scale selection [Lin98] and robust to occlusions due to receptive fields. It produces good results for compact textured objects (see section 5.1) The approach fails completely for objects with sparse texture or objects of small sizes or with holes. The ....

W.T. Freeman and E.H. Adelson. The design and use of steerable filters. Transactions on Pattern Analysis and Machine Intelligence, 13(9):891--906, September 1991.

No context found.

William T. Freeman and Edward H. Adelson. The Design and Use of Steerable Filters. Transactions on Pattern Analysis and Machine Intelligence, vol. 13, no. 9, pp. 891-906, September 1991.

No context found.

William T. Freeman and Edward H. Adelson. The Design and Use of Steerable Filters. Transactions on Pattern Analysis and Machine Intelligence, vol. 13, no. 9, pp. 891-906, September 1991.

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