Increasing measurement accuracy of a chickpea pile weight estimation tool using Moore-neighbor tracing algorithm in sphericity calculation

In this study, a new perspective for increasing the measurement accuracy of a chickpea pile volume and weight estimation tool is proposed. The system proposed uses the principles of machine learning methods and focuses on the adaptation of Moore-neighbor tracing algorithm in sphericity calculation of chickpeas. An experimental setup including a two degrees of freedom moving mechanism, a low-cost laser scanning rangefinder sensor, and a vision unit combining a camera and image processing algorithm is constructed. The methodology proposed is tested to estimate the weight of a chickpea pile using this experimental setup. In order to make comparisons, the estimations are also performed by the use of approaches proposed in the literature. The results of the experimental studies show that at least 5% weight estimation error is obtained when the estimation procedures given in the literature are used. On the other hand, the algorithm proposed in this study yields estimation error less than 0.57%. The details of the procedure proposed, the experimental setup designed and built, the computational environment developed and the experiments conducted are presented in this study.