輪胎是行車安全的必備品，輪胎前端尺寸設計有許多性能與安全考量，使用人工量測尺寸常會有缺乏一致性的問題，而量測儀器則有成本及維護方面的問題。本研究主要是應用影像處理去做輪胎切片尺寸量測，如此解決了在輪胎切片量測時人為與儀器帶來的問題，並在使用類神經網路判定輪胎等級。首先藉由掃描方式擷取影像，由MATLAB撰寫影像處理程式做影像二值化與尋邊處理，簡化影像資訊之後輔以MATLAB撰寫量測尺寸的程式，從影像上尋點並得到尺寸，然後藉由類神經網路中的倒傳遞類神經網路進行等級判別，實驗結果中將得到的尺寸數據做分析研究，分析方式是以設計誤差作為製程能力指標，並驗證類神經網路分類能力，而程式誤差值，則是判定MATLAB程式的準確度的指標，研究設計規劃是以可行性並兼具成本考量來進行，如此得到相對客觀且科學的依據。;We always put tires as the most important issue for safety driving concern. There are safety concern and performance issue in terms of size design of front-end tire. This is an accuracy problem with tire size by manual testing. However, there are cost issue and maintenance problems would be raised which measure by equipment. The main purpose of this study is to measure the size of tire section using image processing system. This method in this study could solve those problems such as manual testing with tire size and cost from the measuring equipment. Additionally, the determination of tire grade uses the artificial neural network. At first, the images select using scan, and image processing program is developed by MATLAB for thresholding and Sobel edge finding. After image information is simplified by image processing program, the program for size measurement is developed by MATLAB. Next, the size acquires by searching the specific position from the image, and the back propagation neural network(BPN) from artificial neural network grades tire size. The data of tire size will be collected from the experiment, design deviation is taken as process capability index, it could verify the capacity of classification based on the artificial neural network. Furthermore, the program deviation is taken as the index to determine the accuracy of MATLAB in this study. The scheme of design in this study proceeds based on cost issue and feasibility assessment, so that the experimental data of this study could meet the expected result.
