| Summary: | 碩士 === 國立成功大學 === 機械工程學系碩博士班 === 97 === The quality of coke greatly influenced the stability, production and coke consumption rate of the blast furnace, the size and uniformity of coke are especially important that should be improved. In this paper, the image processing are used to detect the coke shape and evaluate the coke size.
In this article, the ring light and CCD camera are used to acquire coke image, and we used the image processing such as image enhancement, threshold and morphology to detect the 2D coke shape. We compared the manual painted
shape and the detected shape of cokes by three kinds of threshold method─ Otsu, Entropy and RATS method, the result showed that the average error of area for Otsu method is less than other two, therefore, the coke image segmentation for Otsu method is better than other two. Then, the minimum width of the coke shapes were obtained quickly by rotational searching from the principal axis and the minimum width is defined as the 2D coke size on the image plane. In the static measurement for three size of real cokes─ large(>60mm), middle(30~60mm) and small(<30mm)size coke, the result showed that the coke size of detected shape and manual painted shape are very close, because the average error of coke size between the detected shape and manual painted shape is less than 10%. However, the average error of coke size is more between the
detected shape and 3D shape that reached about 20~30%, since the shape difference between 2D and 3D shape.
In order to reduce the error between the coke size of 2D and 3D coke size,we gather 5 sets of data from large, middle and small coke, respectively, each data include the average difference and the average percentage error of coke size between the detected shape and 3D shape, and used these data to modify the detected coke size in the dynamic measurement. For average difference modification, the results showed that the overall average error of coke size between 2D and 3D shape are about 10.6%, 10.2%, 9.5% and 18.1% for large, middle, small and mixture size coke, respectively, the overall average error are reduced about 22.8%, 12.7%, 18.8% and 12.3%, respectively. For average percentage error modification, the overall average error of coke size between 2D and 3D shape are about 18.1%, 11.2%, 11.5% and 24.5% for large, middle, small and mixture size coke, respectively, the overall average error are reduced about 15.3%, 11.7%, 16.8% and 5.9%, respectively. Generally speaking, the average difference modification is better than the average percentage error
modification.
|
|---|
