The Studying Of Identifying Bars Image

Identifying the No. of bars automatically is widely used in lumbering and iron and steel industry. However, Identifying the No. of bars is confined to circle segmentation for counting of bounded bar. So how to identify the No. of the bounded bar when its segment is not regular can not wait. Now identifying the No. of bounded bar automatically is a hot study, it has the widespread application prospect ant the practical value in this aspect research.This thesis mainly introduces the realization of the section of the bounded bar when picking up image of it on irregular or uneven lighting.In this paper we first recommend the developed trend of identifying the No. of bars, make compare with Template matching,three-point method, and then demonstrate the algorithm being raised in this paper.Then study the pretreatment of the bounded bar, on the base of the research of the foundation imagery processing and theory of Mathematics morphology, to, we use standardization enhance the image, in the follow we make threshold. On the base of the above we do division. Comparing the watershed division and rot, we select watershed to eliminate the stick between each bar. This is a innovation in this thesis. At last of the pretreatment, to thin the section of each bar, using improved OPTA retain template, its function is to less the pixel of the section of each bar and can less the time when identify the No. of the bounded bar.In this paper the mainly study we do is identifying algorithm. In view of the shortcomings of the algorithm that is restrict in sharp on identifying the No. of the bounded bar, we propose two algorithms, one is: in allusion to uneven light section and blurry section of the bar, utilize block image to count the No. of bounded bars. After blocking the image to several blocks, we count the No. of the blocks according the catachrestic of the block. The other algorithm is Identifying the No. of bars Based on two Chain Lists, when scanning the binary image we create two chain lists. At last we determine the No. and the central position of the bar based on the created chain lists.On the base above, we do test on PC on the platform of VC++, enumerate the No. we obtained and do analysis with other algorithm. Test result show: the algorithm we proposed has much higher percentage when identifying the No. of the bounded bars.At last of this paper, we summarize this thesis and make prospect of the future of the study of the identification on bounded bars.