Seed cotton residual film sorting device and algorithm based on hyperspectral imaging and deep learning

Abstract

The invention discloses a seed cotton residual film sorting device and algorithm based on hyperspectral imaging and deep learning; the residual film sorting device comprises a cotton distribution box,a high-spectral imager, a dome halogen lamp, an industrial control computerr, a black rubber crawler belt, a high-speed spraying valve, an air suction opening and a fan. The seed cotton to be treatedenters the cotton distribution box from the cotton inlet to be opened, then the hyper-spectral imager is used for acquiring a reflection spectrum image, and then image processing is carried out through the industrial control computer; after the processing is finished and a certain time is delayed, a driving box drives the high-speed spraying valve to spray the residual film at the corresponding position or the seed cotton containing the residual film to the air suction opening, the fan suck the residual film and the seed cotton, and the residual film located at the bottom of the seed cotton can be sucked into the air suction opening due to the air pressure generated by the fan; and the other seed cotton can fall at the designated position due to inertia, so that the sorting is completed.According to the device and algorithm, rapid and accurate recognition and classification of the seed cotton residual membranes can be realized, and the technical problem of residual film sorting for many years in the industry is solved.

Description

technical field

[0001] The invention relates to the field of identification and sorting of foreign fibers of seed cotton, in particular to a sorting device and algorithm for residual film of seed cotton based on hyperspectral imaging and deep learning. Background technique

[0002] my country is a big cotton production and consumption country, and Xinjiang is the largest and most important cotton planting base in my country. However, Xinjiang adopts large-scale mechanized planting of cotton, and plastic film is inevitably mixed in during the mechanized picking process. The remaining plastic film is not only large in quantity, but also in shape Varies, the color is not too different from cotton, so it is difficult to clean up. At present, many detection algorithms have been used to detect impurities such as cotton bolls, cotton husks, nylon ropes, and cloth strips in cotton. These impurities can be identified by color cameras and black-and-white cameras, but color cameras and ...

Examples

Embodiment Construction

[0021] The present invention will be described in further detail below in conjunction with the accompanying drawings.

[0022] Such as figure 1 As shown, the seed cotton residual film sorting device based on hyperspectral imaging and deep learning in this application has a main structure including cotton distribution box 1, crawler belt 3, dome halogen lamp 4, hyperspectral imager 5, high-speed spray valve 6, and industrial computer (including acquisition card and GPU) 7, drive box 8, air outlet 10, servo motor and encoder 11, fan 12, etc. A cotton box 1 and a dome halogen lamp 4 are arranged above the crawler 3, the dome halogen lamp 4 provides light source for the hyperspectral imager 5, the hyperspectral imager 5 is connected to the industrial computer 7, and the high-speed spray valve 6 is set at one end of the crawler 3, It is also connected with the driving box 8; an air outlet 10 and a fan 12 are set under the high-speed spray valve 6, and the crawler belt 3 is driven ...

technical field

[0001] The invention relates to the field of identification and sorting of foreign fibers of seed cotton, in particular to a sorting device and algorithm for residual film of seed cotton based on hyperspectral imaging and deep learning. Background technique

[0002] my country is a big cotton production and consumption country, and Xinjiang is the largest and most important cotton planting base in my country. However, Xinjiang adopts large-scale mechanized planting of cotton, and plastic film is inevitably mixed in during the mechanized picking process. The remaining plastic film is not only large in quantity, but also in shape Varies, the color is not too different from cotton, so it is difficult to clean up. At present, many detection algorithms have been used to detect impurities such as cotton bolls, cotton husks, nylon ropes, and cloth strips in cotton. These impurities can be identified by color cameras and black-and-white cameras, but color cameras and ...

Examples

Embodiment Construction

[0021] The present invention will be described in further detail below in conjunction with the accompanying drawings.

[0022] Such as figure 1 As shown, the seed cotton residual film sorting device based on hyperspectral imaging and deep learning in this application has a main structure including cotton distribution box 1, crawler belt 3, dome halogen lamp 4, hyperspectral imager 5, high-speed spray valve 6, and industrial computer (including acquisition card and GPU) 7, drive box 8, air outlet 10, servo motor and encoder 11, fan 12, etc. A cotton box 1 and a dome halogen lamp 4 are arranged above the crawler 3, the dome halogen lamp 4 provides light source for the hyperspectral imager 5, the hyperspectral imager 5 is connected to the industrial computer 7, and the high-speed spray valve 6 is set at one end of the crawler 3, It is also connected with the driving box 8; an air outlet 10 and a fan 12 are set under the high-speed spray valve 6, and the crawler belt 3 is driven ...

Images