Objective Cutting occupies a dominant position in mechanical manufacturing, which is the key factor in the development of aerospace, automotive, and electronic industries and other fields. As one of the most important terminals in manufacturing, cutting tool plays an outsized role in removal machining. It is proved that the geometrical parameters have a significant influence on the quality of the workpiece, the efficiency of cutting and the tool life. Given the importance of tool geometry parameters, it is essential to measure these parameters precisely before and during manufacturing. With the increasing requirements of different materials to process, the structures of the cutting tools used to machine become more complex than traditional cutting tools which brings a challenge for the precision measurement. Compared with the common method based on feature images and structured light sensors, the measurement method based on depth from focus presents the advantages of high precision and plenty of data. Hence, a geometry measurement method of cutting tools based on the depth from focus is proposed and the key technique is discussed.
Methods Firstly, to solve the problem that the existing focusing function is not suitable for the sequence images of the measured cutting tools and improve the accuracy of the 3D reduction, an improved focus evaluation function based on the double-threshold Tenengrad function is proposed. Due to the surface irregularities of complex tools, the cutting edge is interrupted, resulting in sharp and diverse edge properties. In order to boost the gray gradient and calculate the edge information that the original convolution operator ignored, the function is improved by adjusting the convolution operator and expanding the direction angle of the edge gradient. Moreover, noise information is taken down using an image preprocessing algorithm and a double threshold constraint to generate a high-quality 3D point cloud. Specifically, Gaussian filtering and the Laplacian image enhancement algorithm are applied to remove the image's natural noise while maintaining all of the image's feature information. The threshold determined by the average value and dispersion degree of the gray gradient is then used to reduce noise from the 3D point cloud. After determining the improved function, the properties of no obvious texture on the tool surface serve to establish the size of the computation window for the function. Secondly, the 3D reduction method of the tool flank is optimized using the image processing algorithm, and a technique of measuring the geometric parameters of the tool flank based on vector arithmetic is proposed. To obtain high-contrast sequence images of the tool surface appropriate for calculation, an image enhancement algorithm based on the adaptive sigmoid function is used. Next, the processed sequence images of the 3D point cloud of the tool flank are produced by using the improved focusing function. Additionally, the cutting tool end face parameters defined by the space measurement plane are described and calculated using the vector angle formula and geometric relationship. Besides, it is necessary to carry out plane fitting on the 3D point cloud of the flank based on the RANSAC (Random Sample Consensus) algorithm. Finally, premised on the previous measurement method, a 3D measurement system is built. Utilizing a ladder formed from standard gauge blocks, the system's depth reduction accuracy is confirmed. Conjointly, a comparative experiment is conducted to measure the geometric parameters of the complex tool end face using a variety of measuring systems and techniques.
Results and Discussions Throughout every experiment that was done, to test the effectiveness of the improved focus evaluation function, different tool surface sequence images are collected, and focus evaluation function curves related to pixels and the whole image are computed by various focus evaluation functions. The results show that the improved focusing function curve is steeper than other function curves intuitively (Fig.4). The sharpness ratio, steepness, sharpness change rate, and local fluctuation are employed to provide a more objective assessment. It also indicates that other functions pale in comparison to the improved focus evaluation function. The average value is 2 082.9%, 5.4%, 6.7%, and 25.7% better than the Tenengrad function, respectively, according to the index order (Tab.1). Furthermore, it is proved that the depth reduction error of the built system is 0.32% (Fig.8). Eventually, the system and measurement method's data for parameter evaluation reveal that the diameter measurement error is less than 3 μm and the apex angle measurement error is less than 0.3° (Tab.3). It is superior to the Tenengrad function's measurements of the angle (less than 1.9°) and diameter (less than 13 μm). In conclusion, the 3D measurement method based on depth from focus can precisely quantify the geometrical characteristics of cutting tools.
Conclusions In this study, an improved double-threshold Tenengrad focusing evaluation function is proposed, which is more suitable for the depth calculation of tool surface sequence images. And the sigmoid function-based image enhancement algorithm is performed to enhance the contrast of sequence images, effectively improving the efficiency and accuracy of the calculation of depth. Further, a prototype of three-dimensional measurement of tool geometric parameters is constructed, and high-quality 3D morphology reconstruction of the tool surface morphology is obtained. Besides, a method for measuring tool geometric parameters based on vector calculation is proposed, and the apex angle and diameter of the tool's inner and outer edges of the main cutting edge are measured. The measured results are within the tolerances of a length measurement error of no more than 10 m and an angle measurement error of no more than 0.5°.
DownLoad: