KR-102963151-B1 - Non-destructive Sweetness Measuring Dvice and Driving Method Thereof

Abstract

The present invention relates to a non-destructive sugar content measuring device and a method for operating the device. A non-destructive sugar content measuring device according to an embodiment of the present invention may include a communication interface unit that receives, respectively, a visible light (RGB) image captured by an RGB camera and an infrared spectral image captured by an infrared camera of a fruit for sugar content measurement, and a control unit that detects an object of the same fruit in the received visible light image and infrared spectral image, detects a defect area within the fruit in the detected object, generates a mask image for the detected defect area, and performs sugar content measurement of the fruit by considering the pixels of the (previously) generated mask image as invalid pixels when applied to an artificial intelligence program.

Inventors

• 김종성
• 전기경
• 권재철

Assignees

• 주식회사 엑시온

Dates

Publication Date

20260511

Application Date

20240104

Claims (7)

1. A communication interface unit that receives, respectively, a visible light (RGB) image captured by an RGB camera and an infrared spectral image captured by an infrared camera of a fruit for sugar content measurement; and A control unit that detects an object of the same fruit in the received visible light image and the infrared spectroscopic image, detects a defect area within the fruit in the detected object, generates a mask image for the detected defect area, and performs sugar content measurement of the fruit by considering the pixels of the generated mask image as invalid pixels when applied to an artificial intelligence program; A non-destructive sugar content measuring device that detects multiple points in a detected fruit object and performs a normalization operation to project and transform the detected fruit object to a designated point using the multiple detected points to remove perspective or rotation.
2. In paragraph 1, The above control unit is a non-destructive sugar content measuring device that inputs the generated mask image into an artificial intelligence deep learning model for sugar content measurement when measuring the sugar content of the fruit, thereby invalidating defective areas of the mask image and training only valid pixels.
3. In paragraph 2, A non-destructive sugar content measuring device, wherein the control unit uses the artificial intelligence deep learning model to generate and utilize a mask image by integrating masks generated for each channel of the visible light or infrared wavelength range through an AND operation when measuring sugar content.
4. In paragraph 3, The above control unit is a non-destructive sugar content measuring device that generates an integrated mask by assigning weights to a specified infrared frequency band when generating the integrated mask image.
5. In paragraph 4, The above control unit is a non-destructive sugar content measuring device that learns a mask image by convolution operation from a mask image region-divided by channel of each wavelength range when performing a learning operation by applying a deep learning model of the above artificial intelligence.
6. delete
7. A communication interface unit receives, respectively, a visible light (RGB) image of a fruit for sugar content measurement captured by an RGB camera and an infrared spectral image captured by an infrared camera; The control unit detects an object of the same fruit in the received visible light image and the infrared spectral image, detects a defect area within the fruit in the detected object, generates a mask image for the detected defect area, and performs a measurement of the sugar content of the fruit by considering the pixels of the generated mask image as invalid pixels when applied to an artificial intelligence program; A step of detecting a plurality of points in the detected fruit object and performing a normalization operation to project the detected fruit object to a designated point using the plurality of detected points to remove perspective or rotation; A method of operating a non-destructive sugar content measuring device, further comprising

Description

Non-destructive Sweetness Measuring Device and Driving Method Thereof The present invention relates to a non-destructive sugar content measuring device and a method for operating the device. More specifically, it relates to a non-destructive sugar content measuring device and a method for operating the device that can minimize mismeasurements caused by impurities or defects (e.g., damage from pests, diseases, etc.) on the skin when measuring the sugar content of fruit based on a non-destructive method using, for instance, an infrared camera. Recently, with the rapid increase in agricultural e-commerce driven by the promotion of agricultural informatization projects, the need for new distribution and logistics technologies is gradually growing, and the necessity for standardization and grading of agricultural products is increasing in a diversified management and distribution environment. However, since fruits exhibit much greater variation in quality and marketability compared to other product categories and are highly susceptible to quality changes depending on the shipping season, efforts are required to provide consumers with a more objective assessment of their quality and marketability. Conventionally, there has been a method for measuring sugar content based on multispectral or hyperspectral imaging of infrared wavelengths (near-infrared and mid-infrared), a NIR-based portable sugar content measuring device, a configuration for reducing errors regarding ambient temperature and light source, and a learning method for measuring sugar content by non-destructive inspection, a method for predicting sugar content by utilizing growth information (e.g., representative sugar content, growing location, number of rainy days, etc.). However, according to conventional technology, when non-destructively measuring sugar content using an infrared camera, the accuracy of the sugar content measurement may be reduced due to defective areas. In particular, there is a problem in that defective areas caused by bruising of the fruit flesh tend to measure high sugar content, while rotten areas tend to measure low sugar content. FIG. 1 is a block diagram illustrating the detailed structure of a non-destructive sugar content measuring device according to an embodiment of the present invention, FIG. 2 is a block diagram illustrating the detailed structure of the normalization unit of FIG. 1, Figure 3 is an example of a box detection result. Figure 4 is an example diagram showing the normalization results of major point detection and projection transformation. Figure 5 is an example diagram showing defect/background region segmentation for each input image, Figure 6 is an example diagram of integrated mask generation, Figure 7 is a diagram illustrating an integrated mask-based convolutional deep neural network structure, Figure 8 is a diagram illustrating a mask learning-based convolutional deep neural network structure, FIG. 9 is a drawing showing a non-destructive sugar content measurement system according to an embodiment of the present invention, FIG. 10 is a block diagram illustrating the detailed structure of the non-destructive sugar content measuring device of FIG. 9, and FIG. 11 is a flowchart showing the operation process of a non-destructive sugar content measuring device according to an embodiment of the present invention. Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram illustrating the detailed structure of a non-destructive sugar content measuring device according to an embodiment of the present invention. As illustrated in FIG. 1, a non-destructive sugar content measuring device (or a (preprocessing) device for measuring sugar content) (100) according to an embodiment of the present invention is a device that performs a preprocessing operation for model learning/inference from an image acquisition unit (111, 121), and includes some or all of an RGB image processing unit (110), an infrared spectral image processing unit (120), a mask generation unit (130), and an infrared spectral image generation unit (140). Here, "including some or all" means that the non-destructive sugar content measuring device (100) may be configured with some components, such as the mask generating unit (130), omitted, or that some components, such as the mask generating unit (130), may be configured by integrating them into the infrared spectroscopic image generating unit (140). To facilitate a sufficient understanding of the invention, it is explained as including all components. The RGB image processing unit (110), infrared spectral image processing unit (120), mask generation unit (130), and infrared spectral image generation unit (140) of FIG. 1 according to an embodiment of the present invention may be configured by a hardware (H/W) module, a software (S/W) module, or a combination thereof. The RGB (or visible light) image processing unit (110) may