UNSUPERVISED TEXTURE SEGMENTATION OF IMAGES USING TUNED MATCHED GABORFILTERS

Recent studies have confirmed that the multichannel Gabor decompositio n represents an excellent tool for image segmentation and boundary det ection. Unfortunately, this approach when used for unsupervised image analysis tasks imposes excessive storage requirements due to the nonor thogonality of the basis functions and is computationally highly deman ding. In this correspondence, we propose a novel method for efficient image analysis that uses tuned matched Gabor filters. The algorithmic determination of the parameters of the Gabor filters is based on the a nalysis of spectral feature contrasts obtained from iterative computat ion of pyramidal Gabor transforms with progressive dyadic decrease of elementary cell sizes. The method requires no a priori knowledge of th e analyzed image so that the analysis is unsupervised. Computer simula tions applied to different classes of textures illustrate the matching property of the tuned Gabor filters derived using our determination a lgorithm. Also, their capability to extract significant image informat ion and thus enable an easy and efficient low-level image analysis wil l be demonstrated.