A new approach to estimate lacunarity of texture images

A new approach is proposed based on gliding-box method to estimate the lacunaritys of texture images. The first approach uses a local threshold to enhance local features of the texture pattern, so that, lacunarity is computed in terms of the local binary patterns that exist in the image, a characteristic which is proven to be a very powerful texture feature. The second approach simply expands the concept of gliding a box over the gray level intensity axis of the texture. As this task is a very time consuming one, we present an algorithm to speed up this process. The results have been tested in a natural texture database, a real and complex problem because the high variability intra-class and great similarity between classes, and compared with traditional lacunarity approaches to show that this new method is computationally simple, convenient and accurate.

• In this study we propose two approaches to estimate the lacunarity of texture images.
• The first approach uses a local threshold to enhance local features of the texture pattern.
• Then, lacunarity is computed in terms of the local binary patterns of the image.
• The second approach simply expands the concept of gliding a box over the gray level intensity axis.
• As this task is a very time consuming one, we present an algorithm to speed up this process.