Ceramic tile texture feature extraction method based on binaryzation and morphology

Abstract

The invention discloses a ceramic tile texture feature extraction method based on binaryzation and morphology. Firstly, median filtering is carried out on an input tile picture and the tile picture is converted into a gray-scale picture; binarizing the grey-scale map by using an Otsu method, and unifying the color of the image to enable the texture feature to be black; secondly, performing connected domain analysis on the picture, removing noisy points outside a feature region to obtain a picture I, performing expansion operation on the picture, performing phase inversion, performing connected domain analysis again, and marking noisy points in the feature region to obtain a picture II; and finally, carrying out bitwise XOR operation on the image I and the image II to obtain a binary image for displaying texture features in black, and carrying out one-time edge detection on the binary image to obtain a sketch representing the texture features of the ceramic tile, so that the extraction of the texture features of the ceramic tile is realized. In addition, the tile texture features can be output in a binary image form or a sketch form, the multiple output forms meet the requirements of different experiments, and a solution is provided for the lack of a data set in the tile texture field.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a method for extracting tile texture features based on binarization and morphology. Background technique [0002] With the rapid development of artificial intelligence, more and more related technologies have been proposed one after another. As the main method to realize artificial intelligence, machine learning needs to predict unknown data based on the characteristics of existing data sets. According to whether the dataset contains label attributes, machine learning can be divided into three categories: supervised learning, unsupervised learning and semi-supervised learning. [0003] The process of machine learning is the process of training a model, and a mature model can well achieve the expected results. However, training a model requires a large number of datasets, and in the tile field, there are very few datasets related to tile texture features, so it is difficult to ca...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem of few texture feature data sets in the field of tiles, the present invention proposes a tile texture feature extraction method based on binarization and morphology, which can quickly extract texture features from a large number of tile image sets, and use two Output the texture features of tiles in the form of value maps or sketches to generate data sets required for machine learning

Images