Intelligent hickory nut shell and kernel sorting machine based on convolutional neural network algorithm

Abstract

The invention relates to an intelligent hickory nut shell and kernel sorting machine based on a convolutional neural network algorithm. The sorting machine comprises a feeding device, a vibrating screen, a rack, an air blowing type sorting device, a power distribution cabinet, a conveying device, an air suction type sorting device, a light shielding plate, an auxiliary light source and an image acquisition system; and the sorting machine further comprises an industrial camera, a lower computer, an upper computer, a data processing system, a target recognition algorithm of a deep learning theory, a regionally-connected-domain segmentation theory positioning algorithm, an air suction type control system and an air blowing type control system. In the design, a plurality of material targets ineach frame of image are positioned on the basis of a regionally-connected-domain segmentation theory, hickory nut materials are not prone to winding, and regionally-connected-domain segmentation is obvious. The upper computer sends center-of-mass coordinates of a selected target to a controller ARDUINO core controller through serial port communication, controls the rapid execution device to reachan actual coordinate system position and accurately controls the suction device to suck away the identified target object. A high-speed air-blowing type execution device is developed to meet variousrequirements.

Description

technical field [0001] The invention relates to a hickory nut shell kernel sorting machine, more precisely, relates to an intelligent hickory nut shell kernel sorting machine with multiple working conditions using a convolutional neural network. Background technique [0002] Pecans are one of the most economically valuable and nutritious nuts with good antioxidant activity. The inner wall of hickory nuts has 2 large partitions and 6-9 small partitions, and the internal structure is complex, with vertical and horizontal ravines. The complex structural features of hickory nuts lead to inconvenient consumption. In recent years, mechanical breaking of shells is commonly used in the production process of hickory nuts. The key issues that need breakthroughs in this process can be attributed to low-damage shell breaking and high recognition rate sorting of broken shells, kernels, and shell-kernel unseparated materials after shell breaking. [0003] The effective separation of ma...

AI Technical Summary

Problems solved by technology

Existing methods for separating walnut flesh from crushed materials using invariant features and a supervised self-organizing map algorithm are very effective for the separation of walnut flesh and shells. There is a certain amount of pollution, and the material treatment of unseparated shells and kernels is not considered

Hickory nuts still contain unseparated individuals after shelling, which would be a waste of resources if not effectively separated

In addition, the recognition rate of unseparated shell and kernel materials based on traditional machine learning algorithms is low, and the sorting result is less than 70%

The current mainstream sorting method has a single feature extraction and requires additional processes to process materials, the recognition rate is not high and the cost is increased

Images