| 摘要 |
PURPOSE:To correctly recognize the furnace wall state by separating a furnace wall image into a joint and a brick face, recognizing the joint state from a joint image, and recognizing the brick face state from a brick face image. CONSTITUTION:The furnace wall image of a coke oven carbonization chamber 10 photographed by a CCD camera 12 at a uniform interval has different density ranges for a joint section and a brick face, and the state of individual furnace walls, particularly damage, cannot be recognized from the density distribution of the whole image. The shading correction, smoothing, and binarization process are applied to the image in advance, then the fact is utilized that the density ranges for the joint section and brick face differ, and a joint section density region 28 and a brick face density region 30 are separated. After the separation, the distribution densities of images of the joint section and brick face are divided into a normal region 32 and a normal/damaged region 34 across a threshold value calculated from the density average value and density dispersion value, and the region 34 contains damage, noise, stuck carbon, and repair trace. Since their density distributions differ, they can be further separated based on the density distributions. |
