Method for automatically identifying field weeds in crop seeding-stage using site and grain characteristic

Abstract

Using position and texture character, the disclosed method in use for identifying farm weed in seedling stage of crop includes following steps: (1) using digital video camera to collect video of seedling stage of drilling crop in farm and seedling of weed on video file; then, through cable, connecting digital video camera to the video collection card inside computer in order to export collected video file and obtaining frame of image form exported video file; (2) using computer to segment between green plant and background of soil; (3) using character of position to identify weed between rows; (4) using texture character to identify weed in rows; (5) weed map between rows of crop, and weed map in rows of crop are added to obtain weed map in farm.

Description

technical field [0001] The invention relates to the technical field of automatic control of precision agricultural machinery and equipment, in particular to a method for automatic identification of field weed information. Background technique [0002] Field weeds compete with crops for water and nutrients, occupy space, affect crop photosynthesis, and interfere with crop growth; if not controlled in time, it will lead to a serious decline in food production. It is estimated that the field weed damage area of ​​only drill crop wheat in my country reaches 10 million hectares, and the annual grain output loss exceeds 4 billion kilograms. [0003] Chemical weed control has become the main method of weed control. In developed countries, the application of herbicides accounts for more than half of the total application of pesticides; in the past 20 years, the application of herbicides in China has increased rapidly, and now the total annual consumption is nearly 470,000 tons. ...

AI Technical Summary

Problems solved by technology

1) Manual monitoring is to use GPS or other positioning systems to manually and automatically record the positions of the weeds seen in the field before the spraying operation, and then draw a weed distribution map. This method has low efficiency, high labor intensity, Relying entirely on manual experience, manual monitoring is powerless in the observation of weed seedlings in large areas

2) Although remote sensing monitoring overcomes many disadvantages of manual monitoring, due to the low spatial and spectral resolution of remote sensing images, it is mainly used to identify several weeds that occur in a large area within a certain growth period, resulting in a relatively low weed recognition rate. Low and low positioning accuracy

Due to the uneven illumination in the field and the similar color of weeds and crops, the automatic identification rate of weeds by color is low, and the automatic identification rate of weeds in this method is usually lower than 80%.

Images