Method for recognizing inline crops and weeds of seedling stage in farmland

Abstract

The invention relates to a method for recognizing inline crops and weeds of seedling stage in farmland, belonging to the technical field of agriculture. The method realizes recognizing of the inline crops and weeds by utilizing a central processing unit system to take left and right images, divide soil background, reestablish three-dimensional images, intercept height information, fit pixel histograms and match recognized crops, wherein the central processing unit is provided with a left camera and a right camera, and receives signals input by the left camera and the right camera. In order to reduce the computational effort of stereo matching to improve the instantaneity of image processing, the invention reasonably and organically combines the division result of step 2 with step 3, i.e. only carrying stereo matching on green objects recognized from the soil background in step 2, omitting soil image parts having a known distance with the cameras, thus greatly reducing the computational effort of stereo matching, obviously quickening the recognition speed and obviously improving the instantaneity of future applications; and the steps after stereo matching effectively ensure the recognition rate.

Description

technical field [0001] The invention relates to an identification method, in particular to an identification method for crops and weeds in rows at the seedling stage of farmland, and belongs to the field of agricultural technology. Background technique [0002] Farmland weeding methods include manual weeding, chemical weeding, mechanical weeding, biological weeding, and weed quarantine. Among them, chemical weeding has obvious effect on sudden weed damage and is widely used at home and abroad. However, if the herbicide is used improperly, it will not only cause phytotoxicity, but also cause excessive pesticide residues in agricultural products, environmental pollution and personnel poisoning. In order to reduce the amount of herbicides and protect the environment, the United States, Europe, Japan, China, etc. have successively carried out research on the precise application of pesticides. Precise pesticide application has become a research hotspot and development trend at ...

AI Technical Summary

Problems solved by technology

However, as previously analyzed, the identification of in-row weeds based on the leaf texture characteristics of crops and weeds has great limitations

Its main shortcoming is, 1) need to set up a database that has stored crops and many weed texture features, then compare the texture features of the field leaves with the texture features in the database, and then judge

This method has a large workload and poor real-time performance; 2) The leaves of crops and weeds grow naturally, and they may not face up, and they may be stained with soil, or overlap each other, or flutter with the wind. These factors will reduce the texture characteristics. recognition rate

[0009] The Chinese invention patent with application number 200510086569.5 discloses a method for automatically identifying weeds in the field. It mainly introduces spraying equipment. The identification of weeds also needs to establish a feature database for comparison, so the workload is heavy and the real-time performance is poor. It has high requirements on image quality and is difficult to apply to farmland sites

Images