Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.
Accurate conversion method of surface bruise size in on-line detection of spherical fruit
Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org
Citation: 2021 ASABE Annual International Virtual Meeting 2100295.(doi:10.13031/aim.202100295)
Authors: Le Liu, Yankun Peng, Long Li
Keywords: Click here to enter keywords and key phrases, separated by commas, with a period at the end
Abstract. Apple is one of the most consumed fruit in the world. With the increasing production of apple, the disadvantages of traditional artificial grading methods, such as time consuming and high cost, are gradually discovered. As a non-destructive apple grading inspection technology, machine vision has the advantages of fast speed and low cost compared with traditional manual grading. But machine vision also has disadvantages in the authenticity of bruise size. Using machine vision to detect bruise in the apple can‘t make a correct classification. Because the bruise in the images can‘t reflect its true size. Therefore, the bruises in the picture need to be corrected before apple detection and classification. Thus, a calculation method was proposed to correct the bruise images. This research mainly includes two parts: the correction of bruise in the horizontal direction and vertical direction. In the horizontal direction, the bruise size correction extracts the coordinates of bruise pixels, regards the cut surface of each pixel row as a circle, and uses the relationship between the coordinates and the central angle to obtain the true size of each row of bruise. In the vertical direction, calculate the distance between pixel rows according to the latitude of the bruise pixel row, establish a trapezoid model for adjacent pixel rows, and calculate the sum of all trapezoid areas as the real bruise area. Finally, in the experiment of the picture, the detection performance of the apple image bruise size correction method is verified. Before the correction method, the R2 of the number of pixels and the actual area is 0.8355, and the root mean square error is 24.53 mm2. After correction, the R2 between the number of pixels and the actual area is 0.9651, and the root mean square error is 10.68 mm2. Compared with the machine vision classification method before correction, the accuracy is improved after using the correction method. The apple bruise correction method that combines horizontal and vertical bruise correction can be used as an effective tool for apple grading. The development of a fast and low-cost apple grading system has great prospects.
(Download PDF) (Export to EndNotes)