Application of digital image processing technique in the microstructure analysis and the machinability investigation

OBRABOTKAMETALLOV Vol. 23 No. 4 2021 TECHNOLOGY These techniques can be classi fi ed into various types based on the type of processing such as point processing, mask processing, noise removal, etc. Based on the input image type, one of the techniques is used. But point processing techniques are most commonly used because of many advantages. In the wide range of applications including microstructural image processing and analysis, digital image processing plays a very important role. A computer algorithm is used for digital image pixel processing. In image processing, complex algorithms are used for easier tasks, which eliminates signal distortion and noise increasing. Two-dimensional images can be modeled for multidimensional systems using digital image processing [14, 15] . In this research work, the same method of digital image processing is employed for the microstructure characterization of fl aked graphite cast iron. There are two methods for capturing the images either using a digital camera or analog. However, effects such as lighting, noise, resolution and others make it necessary to use digital image processing techniques. It makes it possible to convert a relatively poor image into a high-quality one. In the current research work, fl aked graphite images are taken with a digital microscope and further processing has been done for the desired results. The microstructural observation is performed at different magni fi cations as per the requirement. The main purpose of microstructural analysis is to evaluate the microstructure, which is performed to correlate the microstructure as input data with various mechanical properties, including malleability, brittleness and plasticity at the output. [16] Microstructural images with dark spots taken by any method required further post-processing. It consisted of pre-processing, edge detection and fi ltering [17]. Initially, the image is segmented with a pre-de fi ned threshold value with intensifying. After that, the grain boundary of the fl aked graphite in the cast iron is identi fi ed with the edge detection technique. K-Means clustering algorithm The center position of each cluster is de fi ned using K-clusters in the K-means clustering technique [18- 20]. The iterations over the steps continue until a constant minimal sum of square error. The typical steps include calculating the mean value of each cluster, assigning each point to the nearest cluster based on cal- culating the distance from the mean value. At the same time, the following mathematical condition is met. 1 1 , j dK i j j i D G Z = = = - å å where, d j and Z j are the number of pixels and the center of the j th cluster, K is the total number of clusters. The K-means procedure targets to minimize D by satisfying the following condition: 1 . i j j i j g C Z G d Î = å In the dataset G = { g i , I = 1, 2, …, n }, g i is a sample in the d-dimensional space and C = { C 1 , C 2 , …, C q } is the segment that ful fi lled G = Uq i = 1 C i . The microstructure images of grey cast iron with fl aked graphite are evaluated for the machinability. For this purpose, test samples are prepared for analysis. The machining of cast iron workpiece within the fi rst 3.5 mm thickness from the mould-metal interface is a critical problem in hard processing. The microstructure formed within the initial 3.5 mm of the mould-metal interface is evaluated with a digital microscope. The ma- chinability calculation required graphite percentage which is dif fi cult to determine in the digital microscope. The requirement is ful fi lled by k-means clustering in Python . Figure 2 shows the input image fed in the Python program and on the right side processed image is available as an output of the given input image in the plot section of the Python Software . Table 1 shows the percentage of the white area, which is pearlite, and the percentage of the black area, which is graphite, obtained as an output of K -means clustering. The samples are etched in Nital for a clear view of the boundary. Table 2 represents etched condition sample microstructure output. The volume percentage of the pearlite and ferrite is available after etching of the samples.