Among different technique digital image processing is one of the tools that can
well characterize particles shape. Sand being one of the most abundantly found ...
Application of Digital Image Processing for Shape Characterization of Sand Particles Laxman Poudel 1, Bhola Thapa 2, Bim Prasad Shrestha 3, Nabin Kumar Shrestha 4 School of Engineering, Dept. of Mechanical Engineering, Kathmandu University, Dhulikhel, Kavre, Nepal U
U
U
U
U
U
U
U
Abstract- Particles shape is rigorous element to define and is time consuming work through analytical method. A little research has only been explored in shape feature extraction of sediments. So it’s new and difficult to fully characterize sand particles. Among different technique digital image processing is one of the tools that can well characterize particles shape. Sand being one of the most abundantly found and used particles, its shape characteristics are important to define. Different shapes of sand particles are abundant in nature and are most affecting element in hydropower energy generation. It is inherent in irrigation, drinking water pipeline, waste water treatment and many more. A detail understanding of shape feature extraction of sand can bring an important outlet to all people engaged in this field. This part of research work is focused on defining sand particles shape using digital image processing. This discipline is young and is precise to define. Sand particles shape can be defined using geometrical structures which involves mathematics and its derivatives. Shapes are represented by discrete value defined on raster of pixels or edge segmentation and detection and Fourier analysis technique is used to characterize the particles. Shape descriptor is utilized to define exactly the sand particles shape. It describes the region or boundary of sand particles. So study of sand shape using digital image processing with Fourier analysis gives an exact particles shape. This research utilizes nondestructive automation technique online or off-line based on using image processing. Squareness, triangularity, irregularity and elongation are four different measures by which sand particles morphology were characterized developing different sand shape descriptors using complex Fourier analysis. It is revealed that total 21 different shape of sand particles were identified each having different measures. Sand shapes were analyzed and processed using Charge Coupled Devices (CCD) camera, image processing Matrox Imaging Library (MIL) Software and in MatLab 6.5 platform. This new way of vision which cannot be revealed by eye can characterize particles shape easily. Research in shape similarity has a lot of challenges, some solutions and, and for same in use in different application. Keywords: Digital Image Processing, Fourier analysis, Shape descriptor, Sand particles shape
Introduction Shape can be categorized in different ways, but image processing technique is easiest and precise way of classifying. The analysis and effect of shape of sediments can be effectively predicted in some precision level that helps manufacturer of mechanical components of hydropower plant and sediment researchers. It will be a huge asset to improve the efficiency of power plant. This research produces a qualitative and quantitative data to diagnosis the problem inherent to sand problem failure in an automated way which can guide a new vision to design and diagnosis the problem.
It is often very rigorous to define the shape of an object. A little research has only been explored in shape feature extraction of sediments. So it’s new and difficult to fully characterize sand particles. So only a generalized assumption is made to define them. It’s easy to recognize shape visually but very hard to convey the representational or description about shape. Sediments are solid object having 3-dimensional different views. So it’s cumbersome to exactly define all sediment shape types, so only a generalization of distinctly separable shape of particles can be characterized.
Digital Image Technology Technology in today’s world has advanced so faster that using the proper technology can get appropriate result. So as Video technology has taken a great leap, from which we can capture image instantaneously and process to get require output conveniently and affordably. This technology has major advantage in analyzing image more precisely and efficiently and is not a costly technique ao has become a novel approach in this field. Captured image using cameras are recorded digitally which defines in terms of array of pixel in vector space. So the information set by digital image processing helps in defining particles shape, size fraction, and centre of gravity, orientation, count and distribution. Sand Shape feature extraction and measurement is of great concern in sediment and material handling, sediment management, comminuting, constructing and in other various fields. Digital image processing technology has become an efficient and cost effective technology, so shape of the particles has been explores using this techniques with concrete results. The shape of sediment particles directly affects different system. The essence of development of this digital technology has great significance in determining critical coarse shape and size properties: flatness ratio, elongation ratio, flat and elongated ratio, gradation, angularity and surface texture also. This research works includes development of image acquisition system that captures images of sediments instantaneously and in still mode. An ideal image capturing dark room was set up with all image processing software and cameras. Matrox Imaging Library and Magnus Pro software was used to process captured image. Acquisition system was developed to capture Still images in different angles and processed with synchronized Matrox Imaging Library and MagnusPro software for 2-d feature analysis.
Image analysis Sand particles are digitally photographed using CCD camera and analyzed using MIL, Magnus Pro and Matlab image processing software using axial dimensions. Still online image captured at an angle of 450 and 900 and resized the image and background elimination was done to clearly extract all the information of the particle. The image then was converted to gray level image. The image then filtered and segmented to extract the feature in 2 dimensions. The image was processed using different image processing features 1. Particle skeletonization obtained by finding locus or focal reference as a centre to determine the particle geometrical spherical maximal Particle inscribed about its sphere or disc 2. Capture all geometrical edges with polar and rectangular plotting and gridding system. 3. Image enhancement
•
Preprocessing Histogram equalization. • Background elimination was performed. • Taking one slice of stack image, Sobel method was applied for particle’s edge detection and computer codes were developed to find out the center of geometry of that particle. o In each stack of image, a 9 X 9 X 42 sized classifier was allowed to move in such a way that center of classifier coincides with centre of geometry of the particle • The average gray value of this classifier is calculated and a matrix of size 1 X 42 was obtained from one particle. Background elimination and edge detection of images. A slice of image is taken from the stack of image and background elimination was performed by setting all the gray values corresponding to blue color (background) to white.
Sobel method was applied to extract edge of particles which was important to find out the center of geometry of the particles.
Three Dimensional Method; a. Imaging laser slices b. Removing base line and resizing image size c. Converting gray level and then binary image d. Finding edge each of slices e. Merge slices f. Feature extraction Two Dimensional Method; a. Imaging aggregate by 450 angle of CCD1 and 900 angle of CCD2
b. c. d. e. f.
Resize image and subtract background Convert to gray level Noise elimination Segmentation Feature Extraction
Shape of the sand particles are quantized using shape factor which is derived by measuring length of a large particle radii passing through the geometric centre. Mathematically interpreted using root mean square (RMS) deviation of sand particles radii and expressed as
