Classification of texture using random box counting and binarization methods

Abstract

The heterogeneous texture classifications with the complexity of structures provide variety of possibilities in image processing, as an example of the multifractal analysis features. The task of texture analysis is a highly significant field of study in the field of machine vision. Most of the real-life surfaces exhibit textures and an efficiently modelled vision system must have the ability to deal with this variety of surfaces. A considerable number of surfaces maintain a self-similarity quality as well as statistical roughness at different scales. Fractals could provide a great deal of advantages; also, they are popular in the process of modelling these properties in the tasks related to the field of image processing. With two distinct methods, this paper presents classification of texture using random box counting and binarization methods calculate the estimation measures of the fractal dimension BCM. There methods are the banalization and random selecting boxes. The classification of the white blood cells is presented in this paper based on the texture if it is normal or abnormal with the use of a number of various methods.

Detection of lung pathology using the fractal method

“…Fractal analysis can also be used to compare images and detect changes over time. For example, comparing images before and after treatment can help evaluate the effectiveness of treatment and identify possible complications [7].…”

“…Representing the sample initial moment of the q -th order. Then the spectrum of generalized fractal dimensions is calculated by (7),…”

“…𝑞−1 (7) where the function (𝑞) has the form (8). The generalized fractal dimensions of a homogeneous fractal do not depend on q, that is, 𝐷 𝑞 =Dconst.…”

Extensive Review of State-of-the-Art Classification Techniques for Cuneiform Symbol Imaging: Open Issues and Challenges

“…It improves the algorithm's accuracy as well as the rate at which it converges. The default cuckoo search utilizes predetermined values for each of these parameters [13]. Training of ANN using Ant Colony Optimization (ACO): This step involved using ACO to optimize the weight of the neural network.…”

“…Training of ANN using Ant Colony Optimization (ACO): This step involved using ACO to optimize the weight of the neural network. The neural network is trained for pattern classification in the same way that the authors in [13], which trained the ANN using a hybrid technique that included both ACO and BP. BP that got caught up in a local optimal solution.…”