An image processing method for the detection, classification and measurement of pecan cracks
Year: 2025
Type: article
Abstract: This paper describes a software program that detects, classifies, and measures pecan cracks using image processing techniques. Equipped with this software, pecan processors can generate statistics to help monitor the status and efficacy of industrial cracking and shelling processes. The software operates by first detecting the cracks in a pecan shell. The cracks are then classified as longitudinal, circumferential, or open based on their aspect ratio and orientation. Depending on this classification, the software then generates length, circumference, or area measurements. To calibrate and validate the software, a rig is manufactured to rotate the pecans while collecting raw video images. The software is calibrated on a set of 30 pecans and validated on a set of 380. With the validation set, the software accurately identifies 90.6% of cracks, correctly classifies 98.7% of cracks, and measures the length, diameter, and area of cracks within 10% of manual measurements. In addition to aiding pecan processors, the software can facilitate future research on methods for cracking and shelling pecans by providing researchers with an accurate and consistent way of analyzing pecan cracks. • Machine vision software developed to detect, classify, and measure pecan cracks. • Software is successfully validated against manual measurements. • Software performs crack detection on curved surfaces with dark textures. • Software can be used to improve efficiency of post-harvest pecan processing.
Source: Measurement Food
Authors: C.M. Langston, M.W. Jackson, R.B. Davis
Institution: University of Georgia
Language: English
Subfield: Industrial and Manufacturing Engineering
Field: Engineering
Domain: Physical Sciences
SDG: Zero hunger
Open Access status: gold
Estimated APC paid: $1,810
Award: G2709920565
OpenAlex ID: W4415536374