Design And Research Of Wire Diameter Measurement System Based On Image Recognition

As for bar and wire products,quality plays a significant role in the whole safety system.During the rolling process,the diameter size at the outlet of the rolling mill,a key parameter of the rolling process control,is currently detected through the online laser diameter measuring instrument in the finish rolling area in general.However,as the equipment is expensive and hard to be popularized and more than ten rolling mills are required from the primary rolling to the finish rolling for the billet,it is difficult to realize the full coverage of the online detection equipment.Therefore,the close-distance measuring method through the manually handheld caliber is still adopted by many enterprises,which has the disadvantages of the safety hazard and low measuring precision.In the paper,an image-recognition-based diameter measuring device for the portable wire is proposed.With the laser ranging module and combined with the monocular vision ranging theory,the non-contact method is used for measuring the iron-shape size in the vertical direction at a certain distance from the outlet of the rolling mill,thus solving the periodic manual check of primary rolling and intermediate rolling and other iron-shape sizes.The paper emphatically analyzes the image recognition algorithm and researches the improved algorithm applicable for the topic with the realization of the measurement of wire diameter size,of which the detailed researches are shown as follows:As for the impact of all environmental factors in the productions site,including illumination variation,noise,dust and others on the rolled piece image edge,and shining feature of the wire edge,the paper researches several edge detection methods and takes advantage of the color and shape features of the wire to lock the targeted area.Besides,the optimal color channel is selected through the wire image RGB color feature distribution curve.With the improved Zernike moment sub-pixel edge recognition algorithm proposed,the improved gradient filter operator is adopted for the noise sensitivity characteristic of the morphology gradient filter operator firstly to effectively eliminate the noise interference and realize the coarse positioning of the wire edge.Then,for the manual threshold value method with poor precision adopted in the Zernike moment,the partial self-adaptive threshold value method is combined for improvement and the improved Zernike moment is used for secondary positioning of the wire edge to realize the sub-pixel edge detection of the wire.Moreover,the straight line and interference point in other directions of the wire edge is filtered through proposing edge straight-line scanning method with the randomized Hough transform.After this,the boundary can be obtained through the quadratic fit of the point set screened out to determine the boundary pixel distance in the vertical direction.For the image distortion caused by the precision of the original camera lens and camera module deviation,the calibration method of Zhang Zhengyou is deeply researched with the acquisition of the internal reference matrix and distortion coefficient of the vidicon,and the correction of the distortion image.Based on the calibration idea of Zhang Zhengyou,combined with the monocular ranging principle,the iron-shape actual size can be obtained through establishing the space ranging model.Finally,the portable iron-shape size measuring system program is designed,which is composed of the control system module,laser ranging module,camera module,support module,power module and others and equipped with Raspberry Pi4 B embedded system platform.Moreover,it can realize image recognition and processing algorithm based on the Open CV library.