A Study on the On-line Recognition of the Grade of FeO in Sinter by means of Image Recognition Method

摘要

This paper first introduced the measurement methods of ferrous oxide (FeO) in sinter and then proposed a new method for the on-line recognition of the grade of the content of FeO in sinter by means of image recognition method. In the method, a black and white CCD camera equipped with two optical filters, an image collecting card and a personal computer were used to collect the cross-section image of the discharge end of the strand. The image was digitally filtered and segmented. From the segmented binary image, the average area and the average gray level of the sinter pore were abstracted as features. According to the practical situation of sintering production, the content of FeO in sinter was classified to four classes:low, medium, high and too high. The test image samples of the cross-section were collected, the content of FeO associated with every image was determined by the watcher-fire and its membership to each class was claculated. The membership to each class was calcualted. The membership weighted average distance classifing method was used to determine the grade of the content of FeO to the on-lien collected iamge. The measurement principle, the measurement equipment and the experiment result were presented.