Solid wood floor classification method based on deep learning

Abstract

The invention provides a solid wood floor classification method based on deep learning, and the method comprises the steps: carrying out the collection of a floor sample through a visual sensor of anindustrial camera, and building a floor sample data set needed by model training; then, using Gama correction, an Otsu algorithm, median filtering, a local binary mode, an image pixel weighted averagealgorithm and the like for carrying out image preprocessing operation on the acquired image, so that the characteristics of the sample are enhanced, and a large number of operation parameters are reduced for subsequently improving a version VGG16 network; dividing the processed data set into a training set, a verification set and a test set through an algorithm; constructing a VGG16 network model, performing model training after parameters are adjusted, and finally storing weight parameters corresponding to the highest accuracy and the lowest loss value during training to prepare for subsequent unknown floor sample prediction. According to the method, the effect of further improving the efficiency and accuracy of wood floor classification is achieved.

Description

technical field [0001] The invention relates to the classification of wooden floors, in particular to a method for classifying solid wood floors based on deep learning. Background technique [0002] With the increasing development of the national economy, more and more people choose wooden floors for home decoration, which leads to an increasing demand for the number of solid wood floors in my country's wood processing industry. The solid wood floors produced after processing logs are fast and accurate. Classification and packaging have become an urgent problem to be solved. In the current industry, many solid wood floor classification methods still use artificial vision to judge texture and color, and then manually classify. This method is relatively backward and has low efficiency. There are mainly four shortcomings: (1) ) When manually classifying, when the technicians work for a long time, it is easy to produce visual fatigue, which leads to an increase in the classifica...

AI Technical Summary

Problems solved by technology

In the current industry, many solid wood floor classification methods still use artificial vision to judge texture and color, and then manually classify. This method is relatively backward and has low efficiency. There are mainly four shortcomings: (1) ) When manually classifying, when the technicians work for a long time, it is easy to produce visual fatigue, which leads to an increase in the classification error rate; (2) For solid wood floors with similar large textures, but differences in fine textures, artificial vision will be more difficult. It is difficult to distinguish the differences in fine texture features, resulting in classification errors; (3) Different workers have different visions and different understandings of colors during floor classification, resulting in errors in the classification results; (4) Manual classification takes a long time and is inefficient. lower

Images