【摘要】 针对带线缝合针尾孔同轴度检测问题,提出了一种对尾孔放大图像进行边缘检测并拟合的同轴度检测方法。首先根据小波变换原理,利用Db3小波基函数对尾孔放大图像进行了三层小波分解,高频部分利用小波多尺度积边缘检测方法抑制噪声,低频部分利用数学形态学边缘检测方法检测弱边缘信息。然后根据图像信息熵计算了单幅边缘图像的权重系数,进一步将多幅高频边缘图像和低频边缘图像融合,提取到了清晰的尾孔内外圆边缘图像。最后运用最小二乘圆拟合方法得到尾孔内圆与外圆圆心坐标,计算了同轴度。针对0.5 mm缝合针样本的实验结果表明该方法与传统检测方法的同轴度数据差值不超过0.2μm,且具有良好实时性,能够满足尾孔的同轴度检测要求。更多还原

【Abstract】 This paper presents a detection method based on edge detection of tail-hole image and fitting according to the problem of detecting coaxiality of suture needle tail-hole.First,according to wavelet transform theory,several groups of multi-spectral images were decomposed to 3 layers respectively by using Db3 wavelet function.Noise is suppressed by wavelet multi-scale edge detection in high frequency image plane edge extraction.Weak edge information is detected by Mathematical Morphology in low frequency image plane edge extraction.The edge Image fusion of the high frequency image and the low one is gained by calculation of weight coefficient of image entropy.Then,a clear picture of inner and outer tail-hole is produced.Finally,the coaxiality is defined by coordinates of the center of the inner and outer tail-hole,which are obtained on the basis of edge image plane by least-square fitting.By comparing data obtained by the method of detecting coaxiality of suture needle tail-hole provided in this paper with those obtained by traditional method,the 0.5 mm sample experimental results indicate that the different value is less than 0.2 μm.The method introduced in this paper can meet the requirements of coaxiality detection and have a good real-time.更多还原