Israt Jahan, Md. Golam Moazzam, KM Akkas Ali, Mujiba Shaima and Abu Tayeb Muhammad Alimuzzaman
Keywords: Markov random field, machine vision, neighbour boundary, pulse crop, pixel intensity
Abstract: This paper presents a novel approach to analyze the purity of pulse crops by applying machine vision technique. The research concentrated on describing issues related to the development and use of machine vision system for agricultural image interpretation especially for pulse crops. Pulse crops of different stages from different places were collected, saved into computer memory as red, green, blue intensities and converted to Joint Photographic Expert Group format. Four of the most common pulse crops taken from different places were Lentil, Ground Nut, Chick-pea and Split-pea. There were 808 images of pulse crops used for testing and pulse crop purity identification purposes. The success rates of this method for recognized and unrecognized pulse crops of Lentil, Ground Nut, Chick-pea and Split-pea were (84.61, 15.39%), (77.96, 22.04%), (82.19, 17.81%) and (82.69, 17.31%), respectively. Distinct feature of the purity gave the highest percentages of success in analyzing the pulse crop purity.
[View Complete Article]