Title :
Design method for orientation-selective CNN filters
Author_Institution :
Fac. of Electron. & Telecommun., Tech. Univ. of Iasi, Romania
Abstract :
We present a simple design technique for realizing CNN linear filters with an orientation-selective frequency response. This design method relies on a 1-D prototype linear filter (either FIR or IIR), whose shape is identical with the cross-section of the oriented filter to be designed. Considering an IIR realization (with minimum-size templates) for the oriented cosine function and knowing the prototype template coefficients, the oriented filter templates can be easily determined. Simulation results are finally given, to illustrate a task of detecting straight lines with a given orientation from an image, using the CNN oriented filter.
Keywords :
FIR filters; IIR filters; cellular neural nets; 1D prototype linear filter; CNN linear filters; CNN oriented filter; FIR filter; IIR filter; cellular neural nets; frequency response; orientation-selective CNN filters; oriented cosine function; straight line detection; template coefficients; Band pass filters; Cellular neural networks; Design methodology; Finite impulse response filter; Frequency; Gabor filters; IIR filters; Image processing; Nonlinear filters; Prototypes;
Conference_Titel :
Circuits and Systems, 2004. ISCAS '04. Proceedings of the 2004 International Symposium on
Print_ISBN :
0-7803-8251-X
DOI :
10.1109/ISCAS.2004.1328694