This paper presents the multilevel crop image segmentation using a non-local mean two-dimensional (2D) histogram approach. Initially, the 2D histogram is formed with a greyscale and a non-local mean. Moreover, the produced 2D is prone to the slime mould technique when applied with 2D Renyi’s entropy for multilevel segmentation. The performance of the proposed technique is evaluated in terms of fidelity parameters and compared to firefly and beta differential algorithms to evaluate the efficacy. The proposed technique achieved an improvement of 7.37%, 5.97%, 4.52%, and 4.66% of the average peak signal-to-noise ratio with the firefly algorithm and beta differential algorithms at threshold levels 2, 5, 8, and 16, respectively. Correspondingly, the improvement in average root mean square error is 6.25%, 14.70%, 12.96%, and 7.82% at the same threshold levels. Further, the proposed technique is verified with the Raspberry Pi hardware platform. The proposed technique has great potential to be used in the realm of agriculture for image processing applications.