Mango Sorting Mechanism Design and Experimental Research Based on Image Processing

Aiming at the problems of manual sorting and packaging of mango, such as sorting errors, time-consuming and labor-consuming, a mango sorting device was designed based on image processing, and a robust mango surface detection algorithm based on frequency domain filtering was proposed. Firstly, a machine vision image acquisition system is designed to obtain mango images with surface defects. The image is divided into R, G and B color channels, and the R channel is selected for mean filtering, noise reduction and Emphasize enhancement. Then, the preprocessed image is converted into spectrum image by FFT, and then transformed into spatial image by IFFT after removing the components representing texture details, which is different from B color channel to obtain a strong contrast image. Image segmentation based on Otsu method, morphological processing of segmented region, to obtain ideal segmentation results. Analysis results show that the detection accuracy is 84.9%, the accuracy rate is 92.3%, the recall rate is 91.4%, and the F1 score is 0.918. After processing mango images, mango images are classified according to the size of surface defect area, so as to finally realize mango sorting.