TWI490793B - Plant diseases and insect pests identifying method and system thereof - Google Patents

Description

Plant pest and disease identification method and system thereof

The following description relates to an identification method and system thereof, and in particular, to an identification method and system for identifying a disease of a plant and its pests and diseases.

With the concept of green environmental protection becoming more and more popular, more and more people like to plant their own plants. This kind of plant can roughly contain ornamental potted plants or edible plants. When planting plants, the most troublesome thing is that when planting plants, the most troublesome thing is Plants or plants that have been infested by pests have been infected with unknown diseases and gradually withered, withered, and even died.

When the above problems arise, the most effective method is to ask the experts who plant such plants to find ways to control pests and diseases and to effectively cure the plants. However, only a few people have the opportunity to contact and ask. At present, most people are accustomed to using the web search engine to make inquiries, such as Google, Yahoo, etc. However, the search engine must input the correct keywords to effectively respond to the correct search results. Many people find plants when they find pests on the plants. When you are infected with a disease, you often don't know the actual name of the pest or the disease. Therefore, when searching with a search engine, it often fails to find the correct result.

In view of this, the inventors have proposed an identification system and method thereof to improve the above-described drawbacks of the prior art.

The embodiment of the present invention is directed to a method and system for identifying a plant pest and disease, and allows a user to upload a pest image or a plant image to be searched to a cloud server through a mobile device, and the result of the query is returned by the cloud server. In order to easily identify the plant pests and diseases, thereby increasing the convenience of users.

The embodiment of the present invention is directed to a method and system for identifying a plant pest and disease, and the final identification result is obtained by comparing the results of the diseased plant or its pests and diseases by combining multiple identification algorithms. The accuracy of plants or pests is greatly improved.

Based on the above purposes, the present invention provides a method for identifying pests and diseases of a plant, which is suitable for use on a plant that is harmed by a worm or has a disease. The identification method comprises the following steps: 1. Installing an identification management interface on the mobile device. 2. Use the image capture unit to capture the image of the insect body or the leaf image of the plant. 3. Upload the worm image or blade image from the identification management interface to the cloud server. 4. Using a plurality of identification algorithms to compare the worm image or the leaf image with a plurality of worm samples or a plurality of blade samples in the cloud server, so that each of the plurality of identification algorithms generates at least one comparison result. . 5. Generate a comprehensive identification result based on all at least one comparison result, and generate feedback information based on the comprehensive identification result. 6. Return the comprehensive identification results and feedback information to the mobile device. 7. The comprehensive identification result and feedback information are displayed by the identification management interface. The plurality of identification algorithms include using the CIE-Lab color space identification method to detect one of the main colors of the insect image, filtering the background color of the insect image with the main color, and filling the holes in the region of the insect image to perform Comparison of a plurality of worm samples, or a ratio of image correlation coefficients of a plurality of worm samples using a median filter to filter noise of the worm image Correct.

Preferably, the step of generating the comprehensive identification result further comprises: sorting the similarity of the plurality of worm samples or the plurality of blade samples calculated by each of the plurality of identification algorithms, and sorting the plurality of worms after the sorting The volume sample or the plurality of leaf samples are assigned a weight value, and the combined identification value is determined according to the sum of the weight values of each of the plurality of blade samples or the plurality of blade samples in all the at least one comparison result.

Preferably, the plurality of recognition algorithms further comprise an alignment between the worm image and the plurality of worm samples through the shape, the boundary, the texture, the feature or the color.

Preferably, the image capturing unit comprises a micro camera and a built-in camera built in the mobile device, and when the size of the blade of the insect or the plant exceeds a threshold, the built-in camera is used to capture the image of the insect or Leaf image, when the size of the leaf of the worm or plant does not exceed the threshold value, the microscopic camera is used to capture the enlarged image of the worm or the image of the blade.

Preferably, the feedback information includes information about the worm, information about the plant, the manner in which the plant is treated, the manner in which the worm is expelled, or the suggested manner in which the plant is planted.

Based on the above purposes, the present invention further provides an identification system for plant pests and diseases, which is suitable for use on a plant that is harmed by a worm or has a disease, and the identification system comprises an image capturing unit, a mobile device and a cloud server. The image capturing device is used to capture the image of the insect body or the leaf image of the plant. The mobile device includes an identification management interface, and the identification management interface is used to upload the insect image or the blade image and display the received comprehensive identification result and feedback information. The cloud server is configured to receive the insect image or the leaf image from the mobile device, and use the plurality of identification algorithms to respectively input the insect image or the leaf image into the plurality of insect samples or the plurality of blade samples. The row comparison causes each of the plurality of identification algorithms to generate at least one comparison result, and generates a comprehensive identification result according to all the at least one comparison result, and generates feedback information according to the comprehensive identification result, and returns the comprehensive identification result and the feedback information to The mobile device displays the integrated identification result and feedback information in the identification management interface. The plurality of identification algorithms include using the CIE-Lab color space identification method to detect one of the main colors of the insect image, filtering the background color of the insect image with the main color, and filling the holes in the region of the insect image to perform The comparison of a plurality of worm samples, or the use of a median filter to filter the noise of the worm image for comparison with the image correlation coefficients of the plurality of worm samples.

Preferably, the cloud server device sorts the similarity of the plurality of worm samples or the plurality of blade samples calculated by each of the plurality of identification algorithms, and sorts the plurality of worm samples or plurals after the sorting The leaf sample is assigned a weight value, and the combined identification value is determined based on the sum of the weight values of each of the plurality of insect samples or the plurality of blade samples in all of the at least one comparison result.

Preferably, the plurality of recognition algorithms further comprise an alignment between the worm image and the plurality of worm samples through the shape, the boundary, the texture, the feature or the color.

Preferably, the image capturing unit comprises a micro camera and a built-in camera built in the mobile device, and when the size of the blade of the insect or the plant exceeds a threshold, the built-in camera is used to capture the image of the insect or Leaf image, when the size of the leaf of the worm or plant does not exceed the threshold value, the microscopic camera is used to capture the enlarged image of the worm or the image of the blade.

Preferably, the feedback information includes information about the worm, information about the plant, the manner in which the plant is treated, the manner in which the worm is expelled, or the suggested manner in which the plant is planted.

100‧‧‧Advertising System

10‧‧‧Mobile devices

20‧‧‧Cloud Servo

500‧‧‧ Identification system for plant pests and diseases

501‧‧‧ Identification Management Interface

502‧‧‧Image capture unit

503‧‧‧ worm image

504‧‧‧ blade image

55‧‧‧worm body

507‧‧‧ worm sample

508‧‧‧ leaf samples

509‧‧‧ comparison results

510‧‧‧Reward information

511‧‧‧Comprehensive identification results

512‧‧‧ Identification algorithm

S1~S7‧‧‧ steps

The above and other features and advantages of the present invention will become more apparent from the detailed description of the exemplary embodiments illustrated in the accompanying drawings in which: FIG. 1 is an identification system for plant pests and diseases according to an embodiment of the present invention. Block diagram.

Figure 2 is a first schematic view of an identification system for plant pests and diseases according to a second embodiment of the present invention.

Figure 3 is a second schematic diagram of an identification system for plant pests and diseases according to a second embodiment of the present invention.

Fig. 4 is a flow chart showing the steps of the method for identifying plant pests and diseases according to the third embodiment of the present invention.

As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items. When the phrase "at least one of" is preceded by a list of elements, the entire list of elements is modified instead of the individual elements in the list.

Please refer to FIG. 1 , which is a block diagram of an identification system for plant pests and diseases according to an embodiment of the present invention. As shown in FIG. 1, the plant pest and disease identification system 500 includes a mobile device 10, a cloud server device 20, and an image capturing unit 502. The mobile device 10 can include a tablet, a smart phone, or a notebook computer. The cloud server 20 can include a computer host or a server. The image capturing unit 502 can include a built-in camera built into the mobile device 10 or It is a micro camera that magnifies the captured image.

The image capturing unit 502 is configured to capture the worm image 503 of the worm body 55 (not shown) or the blade image 504 of the plant. The mobile device 10 includes an identification management interface 501 for uploading the insect body generated by the image capturing unit 502. The image 503 or the blade image 504, and the identification management interface 501 can further display the received comprehensive identification result 511 and the feedback information 510.

The cloud server 20 can receive the worm image 503 or the blade image 504 from the mobile device 10 through the network, and use the plurality of identification algorithms 512 to store the worm image 503 or the blade image 504 in the cloud server 20 respectively. A plurality of worm samples 507 or a plurality of leaf samples 508 are compared, wherein each recognition algorithm 512 generates at least one alignment result 509, based on all at least one alignment result 509 to generate a comprehensive identification result 511, and according to the synthesis The identification result 511 generates the feedback information 510, and returns the comprehensive identification result 511 and the feedback information 510 to the mobile device 10, so that the comprehensive identification result 511 and the feedback information 510 are displayed on the identification management interface 501.

Please refer to FIG. 2 and FIG. 3, which are a first schematic diagram and a second schematic diagram of an identification system for plant pests and diseases according to a second embodiment of the present invention. As shown in FIG. 2, when the user wants to recognize the worm 55 on the plant, the camera 55 built on the mobile device 10 can take a picture of the worm body 55 on the plant to generate a worm image 503. When the volume of the worm body 55 does not exceed a threshold, the user can additionally use a micro camera to amplify the worm image 503, and then transmit it via the telecommunication network or wireless network in the mobile device 10, which is worth mentioning. Yes, the front view of the worm body 55 is better when it is photographed to facilitate subsequent image recognition processing.

As shown in FIG. 3, the user uploads the worm image 503 to the cloud server device 20 through the identification management interface 501 of the mobile device 10. Before uploading, the user can use the image editing software to advance the worm image 503. The inner wings and tentacles are eliminated to reduce the uncertainty in identification.

After uploading to the cloud server 20, a plurality of recognition algorithms 512 are first used (not shown in the figure). Performing image normalization on the worm image 503, such as performing vertical placement, and then comparing the worm image 503 with the plurality of worm samples 507 for shape, boundary, texture, feature, or color to generate at least one The comparison result 509, for example, one of the recognition algorithms 512 can use the CIE-Lab color space recognition method to detect one of the main colors of the worm image 503, and filter the background color of the worm image 503 with the main color. And filling a hole in the region of the worm image 503 for comparison with a plurality of worm samples 507, thereby generating a list of at least one worm sample 507 conforming to the worm image 503, and another identification algorithm 512 The noise containing the median image 503 is filtered by a median filter to perform an image correlation coefficient with the plurality of worm samples 507. Similarly, the recognition algorithm 512 can also generate the image corresponding to the worm image 503. In addition, at least one of the worm samples 507 is listed, and the cloud server 20 uses the comparison result 509 generated by all the recognition algorithms 512 to generate a comprehensive identification result 511 and its corresponding feedback information. 510.

In detail, the step of generating the comprehensive identification result 511 includes: sorting the similarity similarities of the plurality of worm samples 507 calculated by each of the recognition algorithms 512, and assigning the sorted plurality of worm samples 507 to one. The weight value is further determined based on the sum of the weight values of each of the plurality of worm samples 507 in all the comparison results 509 to determine the comprehensive identification result 511.

For example, if there are two recognition algorithms 512, the shape and texture of the worm image 503 are compared with the worm sample 507, respectively, and the order of similarity according to the shape and the texture is 蚱蜢, muse and螳螂, the 蚱蜢, 螽, and 虫 worm samples 507 are given a weight value of 3, 2, and 1 point, and the other identification algorithm 512 is compared with the worm sample 507 by the color of the worm image 503. If the order of similarity according to color is 蚱蜢, 螳螂, and 螽, then 蚱蜢, 螳螂, and Muse's worm sample 507 is given 3, 2, and 1 points. In this case, only the weight values obtained by 蚱蜢, 螳螂, and 螽 worm samples 507 are added together to obtain {蚱蜢:9,螳螂: 4, Muse: 5} One of the comprehensive identification results 511, the contents of which are shown in Table 1.

The feedback information 510 generated corresponding to the comprehensive identification result 511 may include the profile information of the worm body 55 or the suggested manner of displacing the worm body 55 in this embodiment.

Finally, the comprehensive identification result 511 and the feedback information 510 are transmitted back to the mobile device 10, and displayed on the identification management interface 501. The user can not only know the result of the identification, but also know the most in color, shape and texture. If the integrated identification result 511 is not met, the user can also use the identification management interface 501 to upload a message to the cloud server 20, and then add the cloud server 20 by a professional. The internal worm sample 507 is produced to meet similar identification needs in the future.

In the above embodiment, the worm is taken as an example, and the system can also take the leaves of the plant that produces the disease. For identification, the user can upload the blade image to the cloud server, and after comparing the plurality of blade samples in the cloud server, return the comprehensive recognition result and the feedback information to the mobile device to provide the user with relevant information. Introduction to this plant or treatment.

Please refer to FIG. 4, which is a flow chart showing the steps of the method for identifying plant diseases and pests according to the third embodiment of the present invention. As shown in FIG. 4, step S1 is to install an identification management interface on a mobile device. In step S2, an image capturing unit is used to capture the image of one of the worms or one of the leaves of the plant. For example, in FIG. 2, the image of the worm is captured by a camera of the mobile phone. Step S3 is to upload the insect image or the blade image from the identification management interface to a cloud server. Step S4 compares the worm image or the leaf image with a plurality of worm samples or a plurality of blade samples in the cloud server by using a plurality of identification algorithms, so that each of the plurality of identification algorithms generates at least one comparison. result. Step S5 is to generate a comprehensive identification result according to all the at least one comparison result, and generate a feedback information according to the comprehensive identification result. The comprehensive identification result is shown in Table 1. Step S6 returns the comprehensive identification result and the feedback information to the mobile device. Step S7 displays the comprehensive identification result and the feedback information by the identification management interface.

In summary, the method and system for identifying plant pests and diseases of the present invention provide a function for a user to upload a pest image or a plant image to be searched to a cloud server, and the cloud server returns the result of the query to the mobile device. The purpose of identifying the plant pests and diseases can be easily achieved without directly asking the expert, thereby increasing the convenience of the user. Furthermore, the identification result of the present invention is compared with the results obtained by combining a plurality of identification algorithms. The accuracy of identifying plants or pests is greatly improved.

Although the invention has been particularly shown and described with reference to the exemplary embodiments thereof, It will be apparent to those skilled in the art that various changes in form and detail may be made without departing from the spirit and scope of the invention.

100‧‧‧Advertising System

10‧‧‧Mobile devices

20‧‧‧Cloud Servo

500‧‧‧ Identification system for plant pests and diseases

501‧‧‧ Identification Management Interface

502‧‧‧Image capture unit

503‧‧‧ worm image

504‧‧‧ blade image

507‧‧‧ worm sample

508‧‧‧ leaf samples

509‧‧‧ comparison results

510‧‧‧Reward information

511‧‧‧Comprehensive identification results

512‧‧‧ Identification algorithm

Claims (10)

A method for identifying pests and diseases of a plant is suitable for use on a plant that is harmed by a worm or has a disease. The identification method comprises: installing an identification management interface on a mobile device; using an image capturing unit to capture the worm a worm image or a leaf image of the plant; uploading the worm image or the blade image to the cloud server from the identification management interface; using the plurality of identification algorithms to separate the worm image or the leaf image Comparing with a plurality of worm samples or a plurality of blade samples in the cloud server, such that each of the plurality of identification algorithms generates at least one comparison result; generating a comprehensive identification according to all the at least one comparison result And generating a feedback information according to the comprehensive identification result; returning the comprehensive identification result and the feedback information to the mobile device; and displaying the comprehensive identification result and the feedback information by the identification management interface; wherein the plurality of identification calculations The method includes using the CIE-Lab color space identification method to detect one of the main colors of the insect image, and the main color Color filtering the background color of the image of the worm and filling the holes in the area of the worm image for comparison with the plurality of worm samples, or filtering the noise of the worm image using a median filter To compare the image correlation coefficients with the plurality of worm samples. The method for identifying a plant pest and disease according to claim 1, wherein the step of generating the comprehensive identification result further comprises: counting each of the plurality of identification algorithms Calculating the similarity of the plurality of worm samples or the plurality of blade samples, and assigning the weighted value to the plurality of worm samples or the plurality of blade samples after sorting, and then according to each of the plural numbers The sum of the weight values of the individual worm samples or the plurality of leaf samples in all of the at least one alignment result to determine the comprehensive identification result. The method for identifying a plant pest and disease according to claim 1, wherein the plurality of identification algorithms further comprises transmitting the image of the insect body and the plurality of worm samples through a shape, a boundary, a texture, a feature or a color. Comparison between the two. The method for identifying a plant pest and disease according to claim 1, wherein the image capturing unit comprises a micro camera and a built-in camera built in the mobile device, when the blade or the blade size of the plant When the threshold value is exceeded, the built-in camera is used to capture the image of the insect body or the image of the blade. When the blade size of the insect body or the plant does not exceed the threshold value, the micro camera is used to zoom in and out. The worm image or the image of the blade. The method for identifying a plant pest and disease according to claim 1, wherein the feedback information includes the information of the insect body, the data of the plant, the manner of treating the plant, the manner of expelling the insect body, or the suggestion of planting the plant. the way. An identification system for a plant pest and disease is applied to a plant that is harmed by a worm or has developed a disease. The identification system comprises: an image capturing unit for capturing a worm image of the worm or the plant a blade image; a mobile device comprising an identification management interface for uploading the insect image or the image of the blade and displaying a comprehensive identification result and a feedback information; and a cloud servo a device that receives the image of the worm from the mobile device or the blade Image, and using a plurality of identification algorithms to compare the worm image or the leaf image with a plurality of worm samples or a plurality of blade samples, so that each of the plurality of identification algorithms generates at least one comparison result. And generating the comprehensive identification result according to the at least one comparison result, and generating the feedback information according to the comprehensive identification result, and transmitting the comprehensive identification result and the feedback information to the mobile device, so that the comprehensive identification result and the feedback information are obtained Displayed in the identification management interface; wherein the plurality of identification algorithms include using a CIE-Lab color space identification method to detect a main color of the insect image, filtering the background color of the insect image with the main color, and filling a hole in the region of the worm image for comparison with the plurality of worm samples, or a noise filtering the worm image using a median filter for image association with the plurality of worm samples The ratio of the coefficients. The identification system for plant pests and diseases according to claim 6, wherein the cloud servo device compares the plurality of insect samples or the plurality of leaf samples calculated by each of the plurality of identification algorithms. Degrees are sorted, and the plurality of worm samples or the plurality of leaf samples after sorting are assigned a weight value, and then at least one comparison result is obtained according to each of the plurality of worm samples or the plurality of leaf samples. The sum of the weight values within the decision to determine the comprehensive identification result. The identification system for plant pests and diseases according to claim 6, wherein the plurality of identification algorithms further comprises transmitting the insect image and the plurality of worm samples through a shape, a boundary, a texture, a feature or a color. Comparison between the two. The identification system for plant pests and diseases according to claim 6, wherein the image capturing unit comprises a micro camera and a built-in camera built in the mobile device, when the size of the blade or the blade of the plant When the threshold value is exceeded, the built-in camera is used to capture the image of the insect body or the image of the blade. When the blade size of the insect body or the plant does not exceed the threshold value, the micro camera is used to zoom in and out. The A worm image or an image of the blade. The identification system for plant pests and diseases described in claim 6 wherein the feedback information includes information about the insect body, information about the plant, the manner of treating the plant, the manner of expelling the insect body, or the suggestion of planting the plant. the way.