Laser-enhanced seed cotton residual film on-line detection system

Abstract

The invention relates to a laser-enhanced seed cotton residual film online detection system. Enhanced line laser is used as a light source, a seed cotton residual film image is acquired, then mean filtering and SLIC clustering segmentation processing are performed on the image, and an initially marked clustering segmentation image is generated. Texture features are extracted such as relevancy, contrast ratio, energy, homogeneity and entropy of a super-pixel region of the initial segmentation image as feature values, region attributes of seed cotton and residual films are marked, and an SVM classifier is trained; and the superpixel blocks of the same category in the measured image are combined by using an SVM classifier, thereby realizing segmentation of the seed cotton residual film image,calculating the pixel areas of the seed cotton and the residual film, estimating the content of the residual film in the seed cotton according to the pixel area of the residual film, and realizing online evaluation of the content of the residual film in the seed cotton.

Description

technical field [0001] The invention relates to an evaluation system for residual film content in seed cotton processing, in particular to a texture feature-based SVM residual film recognition system for seed cotton, which belongs to the field of agricultural engineering. Background technique [0002] Cotton is an important production material and has an important impact on national economic and social development. In the cotton harvesting process, the operation of the cotton picker improves the work efficiency, and it is inevitable to mix impurities such as residual film. The material properties of residual film and cotton are similar, resulting in a low detection rate of residual film. On the one hand, the residual film affects the quality grade of cotton, and on the other hand, it also affects the processing techniques such as spinning and dyeing, thus affecting the quality of woven products. [0003] At present, the foreign fiber cleaning technology in lint is relative...

AI Technical Summary

Problems solved by technology

CN201910800155.6 discloses a hyperspectral visualization labeling algorithm for seed cotton residual film for deep learning, which applies hyperspectral technology to the field of seed cotton residual film identification, and collects hyperspectral images of seed cotton residual film at 1000-2500nm through a hyperspectral imager , and then identify and classify the residual film, but the cost is high and it is not suitable for cotton spinning enterprises

Images