Two New Methods of Boundary Correction for Classifying Textural Images

With the growth of technology, supervising systems are increasingly replacing humans in military, transportation, medical, spatial, and other industries. Among these systems are machine vision systems which are based on image processing and analysis. One of the important tasks of image processing is classification of images into desirable categories for the identification of objects or their specific areas. One of the common methods is using an edge finder in image classification. Due to the lack of definite edges in many images obtained from various sciences and industries such as textural images, the topic of textural image classification has recently become of interest in the science of machine vision. Thus, in this article, two methods are proposed to detect edges and eliminate blocks with non-connected classes based on fuzzy theory and weighted voting concepts in classifying textural images. In the proposed methods, the boundaries are corrected using fuzzy theory and weighted voting concepts. Using the proposed methods can help improve the definition of boundaries and classification accuracy.