【摘要】 针对高炉料面红外图像特征难以准确提取的问题,提出一种基于多源信息融合和免疫遗传算法的最大模糊熵分割方法。根据专家经验和多源过程检测信息,将高炉料面图像分为高温和低温子图像,采用免疫遗传算法和最大模糊熵分别对子图像进行分割,再将分割后的图像融合。采用国内某钢铁公司高炉炉顶摄像机拍摄的图像进行图像分割比较实验。研究结果表明:该方法有效地利用多源信息和高低温图像特征,充分发挥最大熵法分割精度高和受目标大小影响小等优点,通过免疫疫苗接种提高遗传算法搜索最佳阈值的收敛速度;该方法能高效准确地提取高炉料面温度特征,及时发现高炉异常炉况,更符合工业实际应用要求。更多还原

【Abstract】 Considering the fact that the fact that the feature of image for burden surface temperature profile in BF can not be accurately extracted,a kind of maximum fuzzy-entropy image segmentation method based on the multi-source information fusion and the immune genetic algorithm for burden surface infrared image was proposed.Firstly,according to the expert experience and the multi-source detected information of the BF,the image of burden surface was divided into two sub-images,named as high or low temperature sub-image.Then the two sub-images were segmented respectively by using the maximum fuzzy-entropy and the immune genetic algorithm.Finally,the two segmented sub-images were fused into one image.The results show that since both the multi-source information and the temperature feature of the image are thoroughly taken into account,this method makes full use of the advantages of maximum fuzzy-entropy with the high precision and less influence from the target size,and by adopting the immune algorithm,it also improves the convergence speed of genetic algorithm for searching optimal segmentation threshold.The method is effective and accurate to obtain its temperature feature and more timely to discover its abnormal conditions.It is proved to be a better method to meet the demands of the industrial application.更多还原