Living standing tree rapid identification method based on hybrid multi-scale feature model

Abstract

The invention provides a living standing tree rapid identification method based on a hybrid multi-scale feature model, belongs to the field of detection, and solves the problems in the prior art. The living standing tree rapid identification method based on the hybrid multi-scale feature model comprises the steps that the visible light image and the infrared thermal image of a target area are acquired; feature level image fusion is performed on the acquired visible light image and the infrared thermal image by using a live wire + contourlet algorithm; the directional gradient histogram of a living standing tree target area is established for the fused feature level image; image features are extracted on different scales of the directional gradient histogram so that the hybrid multi-scale feature model is established; the hybrid multi-scale feature model acts as a positive sample, and the target living standing tree in the feature level image is identified; and the identified target living standing tree is marked and saved in the feature level image. The method has the advantages that the forestry environment can be rapidly identified, the operation targets can be distinguished, the operation efficiency can be enhanced and the operation safety can be guaranteed.

Description

technical field [0001] The invention belongs to the detection field, and relates to a standing tree recognition method, in particular to a fast standing tree recognition method based on a mixed multi-scale feature model. Background technique [0002] Forest tending operations are mainly forest management activities such as tending and thinning, fixed tree and branch pruning, weeding, cutting and irrigation, construction and maintenance of simple roads in the tending area, collection and transportation of tending materials, treatment of tending residues, and forest land clearing. The forestry machinery used includes brush cutters, harvesters, delimbers, skidders, etc. When these machines perform tending operations, due to the very complicated operating environment in the forest area, the operator cannot accurately distinguish the operating objects only by visual inspection. The operating efficiency of forestry tending machinery has been greatly reduced, and domestic forestry ...

AI Technical Summary

Problems solved by technology

[0005] 1. At present, the research on forest tree measurement is mainly based on the measurement of individual standing tree parameters and three-dimensional reconstruction, while the research on the real-time feedback method of visual operating environment information for forestry machinery equipment is still relatively rare;

[0006] 2. Poor efficiency and accuracy of tree measurement research

The commonly used 3D laser scanner can directly obtain the terrain information of the environment. However, the large amount of point cloud data obtained from the surface of the object not only increases the system load, but also greatly reduces the efficiency of subsequent processing, which is far from meeting the requirements of the mechanical equipment used by people. Judgment needs for environmental perception and recognition

Methods such as binocular vision detection have low recognition accuracy due to the mismatching phenomenon caused by blurred target features, and cannot meet the requirements of forestry machinery and equipment for operation accuracy and efficiency in practical applications;

[0007] 3. Artificial intelligence technology has not been deeply applied in the field of tree measurement

Compared with developed countries, due to the relatively late application of computer technology in the forestry field in my country, the scope of application is relatively small, which directly leads to the relatively low level of overall automation and informatization of forestry

Images