CN111291702A

CN111291702A - Identification method for distinguishing locust species by image recognition technology

Info

Publication number: CN111291702A
Application number: CN202010105669.2A
Authority: CN
Inventors: 张昕然
Original assignee: Beijing Garden Bio Technology LLC
Current assignee: Beijing Garden Bio Technology LLC
Priority date: 2020-02-20
Filing date: 2020-02-20
Publication date: 2020-06-16

Abstract

The invention provides an identification method for distinguishing locust species by using an image identification technology, which summarizes different biological characteristic identification methods of locust species, such as wing-foot appearance characteristics, body length and the like, and an image identification algorithm engineer can train an identification model according to the identification method, so that locust species, such as Asian dolly locust, migratory locust, crinkled knee locust, rice locust and the like, are automatically judged and identified from images of the locust, thereby monitoring occurrence species and occurrence density of different locust, effectively meeting the requirement of locust monitoring work and enabling the locust to be more effectively prevented and controlled. The method can accurately and efficiently distinguish the locusts of different types, has the characteristics of high efficiency, simplicity, rapidness, automation, strong specificity, high accuracy and stability and the like, and can save a large amount of manpower and material resource cost.

Description

Identification method for distinguishing locust species by image recognition technology

Technical Field

The invention relates to a biological research technology of locust in the field of plant protection.

Background

The locust is the most serious pest which is historically harmful to crops in China, the occurrence density of field locusts can be automatically monitored by utilizing an image recognition technology at present so as to judge the occurrence situation of the locusts, but the species of the locusts cannot be accurately identified so as to monitor what the species of the locusts occur and the occurrence density of locusts of different species, and the species of the locusts can be accurately judged in the locust plant protection work at present, so that the key point and the key point of actual locust monitoring can be realized, the current situation and the development trend of the locusts can be accurately positioned, and more accurate information and a precedent are provided for timely taking control measures.

Disclosure of Invention

Aiming at the problems, the invention provides an identification method for distinguishing locust species by using an image identification technology, which summarizes different biological characteristic identification methods of the locust species, such as wing-foot appearance characteristics, body length and the like, an image identification algorithm engineer can train an identification model according to the identification method, so that the locust species, such as Asian dolly locust, migratory locust, crinkled knee locust, rice locust and the like, can be automatically judged and identified from an image of the locust, the occurrence species and the occurrence density of different locust can be monitored, the locust monitoring work requirement can be effectively met, and the locust can be effectively prevented and controlled.

The invention adopts the technical scheme to realize the purpose, and has the following advantages: 1. according to years of researches on the behavior of the locust, the invention summarizes different biological characteristics of several common locust, such as appearance characteristics, body length and the like of the wings and feet, and can accurately and efficiently distinguish the different types of locust. 2. The locust identification method is used for carrying out image identification and distinguishing on the locust, has the characteristics of high efficiency, simplicity, rapidness, automation, strong specificity, high accuracy and stability and the like, and can save a large amount of manpower and material resource costs.

Drawings

FIG. 1. Asian locusts

FIG. 2. migratory locust

FIG. 3 locust with crinkled knees

FIG. 4 shows locusta oryzae

FIG. 5 locust image recognition device

FIG. 6 photo of locust

The specific implementation mode is as follows:

the present invention will be described in detail below with reference to the accompanying drawings.

As shown in fig. 1, the body length appearance of the identifying method of the acridid in asian dolly is characterized by small size and medium size, and the body length: 18.5-37.0 mm. The appearance of the fin is characterized in that: the front wing is farther beyond the end of the hindfoot femoral condyle. The anterior wing middle insertion vessels are obvious, and the superior sound teeth are developed; the basal half had 2 large dark spots. The middle part of the back wing has a dark transverse band, and the base part is light green.

As shown in figure 2, the migratory locust identification method has the characteristics of large body length and appearance, and large body length; 32.4mm-52.8 mm. The appearance of the fin is characterized in that: the front wings are developed and far exceed the end parts of the hind foot femoral joints; the middle leap vein in the middle pulse region is obvious, and near the anterior elbow, there are obvious sounding teeth on it, and many small transverse veins arranged in parallel are grown near it. The base of the hind wing is light yellow. The middle ridge on the upper side of the femoral head has obvious saw-toothed shape and is provided with thorns at the outer ends.

As shown in fig. 3, the method for identifying locusta crinis has the characteristics of medium body length, body length: 23-32 mm. The appearance of the fin is characterized in that: the anterior wing has dense and fine brown spots, and the central region of the anterior wing has obvious central insertion veins. The posterior wing is shorter than the anterior wing, the front edge is bent in an S shape, the main longitudinal pulse is obviously thickened, and the base is rose-red and transparent.

As shown in FIG. 4, the method for identifying locusts has a body length of 15.1-40.5mm and a medium body size. The appearance of the fin is characterized in that: the front and the back are developed, the front wing is longer and just exceeds the middle part of the shank of the hind paw, and the back wing is slightly shorter; the front wing is green, or the front edge is green, the rest is brown, and the back wing is colorless; the radial vessels of the anterior wing have small transverse vessels arranged irregularly.

Experimental example 1:

in this embodiment, as shown in fig. 5, the image recognition device is placed on a white board, various common locusts are placed on the white board, and by taking a picture, as shown in fig. 6, the body length and wing features of the locusts can be clearly judged by naked eyes, so that it can be seen that by image recognition algorithm training, the species of the locusts can be clearly distinguished by taking pictures of several kinds of locust identification features summarized by the present invention. Thereby accurately judging the development status and trend of the locust.

The present invention has been described in terms of the above embodiments, and equivalents thereof based on the principles of the invention are not excluded from the scope of the invention.

Claims

1. An identification method for distinguishing locust species by using an image recognition technology is characterized by comprising the following steps: and summarizing different biological characteristic identification methods of the locusts, such as appearance characteristics, body length and the like of the wings, and an image recognition algorithm engineer can train a recognition model according to the identification method, so that the locusts, such as Asian dolly locusts, migratory locusts, crinkled knee locusts, rice locusts and the like, can be automatically judged and recognized from the image of the locusts, and the occurrence types and the occurrence densities of different locusts can be monitored.

2. The Asian locusta biological characteristic identification method according to claim 1, characterized in that: the Asiatic locusts have the length and the appearance characteristics that the size is medium and small, and the length is as follows: 18.5-37.0mm, the appearance characteristics of the fin are as follows: the front wing is farther beyond the end of the hindfoot femoral condyle. The anterior wing middle insertion vessels are obvious, and the superior sound teeth are developed; the half part of the base has 2 large dark spots, the middle part of the back wing has a dark transverse band, and the base is light green.

3. The method for identifying the biological characteristics of migratory locust as claimed in claim 1, wherein: the migratory locust has the advantages of large body length and appearance characteristic; 32.4mm-52.8mm, the appearance characteristics of the fin are as follows: the front wings are developed and far exceed the end parts of the hind foot femoral joints; the middle leap vein in the middle pulse region is obvious, and near the anterior elbow, there are obvious sounding teeth on it, and many small transverse veins arranged in parallel are grown near it. The base of the hind wing is light yellow, the middle ridge on the upper side of the femoral head is obviously jagged, and the outer end of the hind wing is not stabbed.

4. The biological feature identification method for the migratory locust of the knee joint as claimed in claim 1, wherein: the body length of the locusta crinkled by knees is characterized by medium body type and long body: 23-32mm, the appearance characteristics of the fin are as follows: the anterior wing has dense and fine brown spots, the median pulse region of the anterior wing has obvious intercalated veins, the posterior wing is shorter than the anterior wing, the front edge is bent in an S shape, the main longitudinal veins are obviously thickened, and the base is rose red and transparent.

5. The method for identifying biological characteristics of rice locust as claimed in claim 1, wherein: the body length and the appearance characteristics of the locusts are medium body type, the body length is 15.1-40.5mm, and the appearance characteristics of the wings are as follows: the front and the back are developed, the front wing is longer and just exceeds the middle part of the shank of the hind paw, and the back wing is slightly shorter; the front wing is green, or the front edge is green, the rest is brown, and the back wing is colorless; the radial vessels of the anterior wing have small transverse vessels arranged irregularly.