Agricultural Research Management Information System

Home
Research Summary
Search
About Us
- ARMIS
- Brochure
Contact Us
Report
User Request
Data Input
Help
- Operation Manual
  - PDF
  - Video
- Program Area & Commodity
We have reached 37600 number of research entries at this moment.

- Logout

Research Detail

Home
Research
Detail

Detection and Contouring of BAU-kul Using Image Processing Techniques

MM Rahman
Department of Agricultural Engineering, Bangabandhu Sheikh Mujibur Rahman Agricultural University, Gazipur 1706, Bangladesh

MMH Oliver
Department of Agricultural Engineering, Bangabandhu Sheikh Mujibur Rahman Agricultural University, Gazipur 1706, Bangladesh

Abstract/ Executive Summary:

Automated grading and sorting of fruits during the harvesting period are needed for securing better market prices. In order to introduce such automation facilities in Bangladesh, edging and contouring information of the locally grown fruits is important. This study reports the first endeavor towards the use of image processing techniques for a popular jujube variety (BAU-Kul) in Bangladesh. Image processing techniques were used for segmentation, and contouring on the basis of color Thresholding, edge detection and contour detection in Python-OpenCV software. Six random samples of BAU-Kul fruit were used for the research. Perimeter lengths obtained from the image analysis of the six samples ranged from 17.9 cm to 20.20 cm with an average of 19.29 (±1.02) cm. The measured lengths, on the other hand, varied from 16.2 cm to 19.1 cm with an average of 17.75 (±1.3) cm. Consequently, the average error in calculation was limited to only 7.98%. This indicates the fact that images captured through mobile devices can be used for detection and contouring of BAU-Kul samples with fairly high accuracy (92.02%). This information provides a foreground basis of automation for the grading and sorting systems of BAU-Kul fruits in Bangladesh.

Keywords: BAU-Kul, Contour detection, Edge detection, Image segmentation

Location: Department of Agricultural Engineering, Bangabandhu Sheikh Mujibur Rahman Agricultural University, Gazipur 1706, Bangladesh

Start Date:

End Date:

Program Area: Postharvest and Agro-processing

Commodity: Jujube

Objectives:

To shed some light on this area by employing color segmentation and contour detection approaches for BAU-Kul images.

Methodology:

The overall process of the BAU-Kul detection system has been illustrated in the following flowchart. Theoretically, the process of BAU-Kul detection system involves low- level processing and high level processing. In low-level processing, the digital color images were captured using a portable mobile devices. The captured RGB images were converted to HSV, and later to grayscale images in order to extract the Thresholding images. In high level processing, the image was then processed through the Canny Edge function in order to obtain the parametric shapes of BAU-Kul samples. At the final step, the Thresholding and Canny Edge images were used to obtain the desired fits of contours around the BAU-Kul samples. The implementation of an algorithm for contour fitting of BAU-Kul therefore, comprises of the following consecutive steps i.e., (a) image acquisition (b) pre-processing (c) thresholding followed by canny edge detection and (d) contouring. These steps and their mathematical models have been described as follows: Samples of BAU-Kul were collected from the local market in Dhaka city in the month of February 2019. Six randomly collected BAU-Kul samples were pictured using mobile devices. The samples were laid out on a non-reflective surface, and naturally diffused sunlight (2-5 W/m²) was used for capturing these images. The device was set a fixed height so as to keep the focal length within (26-33 mm) for all the images so that the shadow effects could be minimized.

Source of Information: Ann. Bangladesh Agric. (2019) 23(2) : 15-25

Article Link (Online Journal): DOI: https://doi.org/10.3329/aba.v23i2.50052

Funding Source:

Budget:

Total Budget (Tk.) :

Achievement/Findings:

Image processing techniques are used to measure the qualities of fruits based on color space model, Thresholding, edge and contour detection methods. In this research, Thresholding segmentation, edge and contour detection were obtained from the color images of BAU-Kul by using Python-OpenCV platform. The edge and contour of BAU- Kul images were found to be good enough to be used for operational purposes. The contours will however, be more accurate if the suitable lighting conditions are employed in order to minimize the shadow of images on the background surfaces. Nonetheless, the algorithms employed in this manuscript can be used for determining the BAU- Kul samples with considerable accuracy (92.02%). The information generated from this research will be helpful for commercial grading and sorting facilities thriving for automation. This would ultimately help the growers get better price according to the shape and sizes of BAU-Kul.

Knowledge Management: Journal