Shape Identification and Particles Size Distribution From Basic Shape Parameter Using ImageJ

           The researchers at Mississippi University developed an ImageJ plug-in that extracts the dimensions from any digital array of particles soon after identifying their shapes and determining their particle size distribution. The paper describes the plug-in's development and its application to food grains and ground biomass. It is discovered that the plug-in was applied successfully to analyze the dimensions and size distributions of food grains and ground Miscanthus particle images.
          This research deals with three main objectives: the development of ImageJ plug-ins, determining the effects if shape, size, and orientation, and demonstrating an application to samples. Plug-ins are useful because when trying to achieve your output manually, it can be very time consuming and crosses the line of user subjectivity. The computer vision method determined the distribution and amount of garlic, parsley, and vegetable ingredients in pasteurized cheese with an accuracy of over 88%, compared wth the sensory method. However, there are multiple methods used because quick and accurate particle size distribution analysis is most desirable; especially when dealing with granular or particle materials. With ImageJ one can map the actual particle to an equivalent ellipse and perimeter match. The study establishes that fitted ellipse dimensions produce relativity good estimates.
           The results and discussion explained multiple plug-ins that were tested. First, they deliberated that shape-based corrections factors that fit ellipse dimensions are essential for measuring linear dimensions. Fitted ellipses dimension better than bounding rectangles because it produces less deviation. It was then noted that for triangle shapes, the FMR forms other shapes for easy identification. These findings of shape dimensions leads to the development of shape identification strategy. Later, an image of a known geometric shape and measurement was drawn and tested. However, some miscalculations occurred with samples. It was found that to avoid misclassification further research would need to be conducted.
           The effect of particle shapes with getting the measurements in length and width was calculated using geometric reference particles through absolute deviation. It was concluded that the shape does not have any effect on mean absolute deviation, and the area was inversely proportional to the mean absolute deviation. They found that their results indicated good shape classification of the plug-in only with round particles. It was also observed a coincidence of arithmetic and geometric mean lengths. The normal distribution shaped curve was seen towards the left, allowing an increase since the length was so small. The final plug-in that was tested was a drawback of image processing method. However, it can be used only with separation of a mechanical system.
           The research conducted shows the great potential of ImageJ plug-ins proving to be efficient and reliable. They are ready to give solutions in specific needs to machine vision applications.