Stereo Imaging Depth Map for Plant Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional farming systems face inefficiencies in treating individual plants within a field, often resulting in waste and high labor costs due to the inability to accurately distinguish between crops and weeds, and existing imaging technologies are limited in their ability to identify and treat plants at the individual level.
Innovation Solution
A farming machine equipped with a pair of image sensors, one producing color images and the other multispectral images, generates a stereo image pair to create a depth map, allowing for accurate identification and targeted treatment of crops and weeds using machine learning and the Normalized Difference Vegetation Index (NDVI), enabling precise adjustments for treatment mechanisms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional imaging systems (satellite, color, thermal) are used to identify plants, then the system complexity is low, but the measurement precision for individual plant identification is poor
Solution Approach 1:
The patent combines multiple imaging modalities (color camera and multispectral infrared camera) into a single integrated system. The color camera captures visible light information for basic plant detection, while the multispectral infrared camera captures thermal and near-infrared information for enhanced plant discrimination. By merging these complementary imaging systems and processing their data together, the patent achieves high-resolution individual plant identification that neither system could achieve alone, resolving the contradiction between measurement precision and device complexity.
Solution Approach 2:
The patent transitions from two-dimensional color imaging to three-dimensional stereo imaging by incorporating depth information through infrared thermal imaging. The multispectral camera provides additional spectral dimensions (thermal infrared band) that enable depth perception and three-dimensional plant characterization. This dimensional enhancement allows precise individual plant identification and spatial mapping, achieving high measurement precision while the integrated processing maintains manageable system complexity.
2Loss of substance
If conventional spray treatment systems are used to treat all plants in a field, then the productivity is high, but the loss of substance increases due to treatment waste
Solution Approach 1:
The patent implements local quality by transitioning from uniform field-wide treatment to individualized plant-specific treatment. The integrated imaging system identifies and characterizes each plant individually, allowing the treatment system to apply different treatments (fertilizer, herbicide, water) to different plants based on their specific needs. This selective approach eliminates treatment waste on healthy plants while ensuring adequate treatment for plants that need it, thereby reducing substance loss without compromising overall productivity.
Solution Approach 2:
The patent incorporates real-time feedback through the imaging system that continuously monitors plant health status, species identification, and spatial location. This feedback information is processed to dynamically control the treatment application system, adjusting treatment type and dosage for each individual plant based on current conditions. The closed-loop feedback mechanism ensures treatment is applied only where and when needed, minimizing waste while maintaining high treatment efficiency and productivity.
3Loss of substance
If manual treatment application is used to treat individual plants, then the loss of substance decreases, but the productivity decreases due to labor intensity
Solution Approach 1:
The patent replaces manual mechanical treatment application with an automated system that integrates imaging, processing, and treatment delivery. The color and multispectral cameras automatically capture plant information, the processing system identifies and characterizes individual plants, and the treatment system automatically applies appropriate treatments. This automation eliminates labor-intensive manual operations while maintaining precise individual plant treatment, thereby improving productivity without sacrificing treatment precision and reducing substance loss.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system allows for efficient and accurate identification and treatment of individual plants, reducing waste and labor costs by using a combination of color and multispectral imaging to differentiate between crops and weeds, thereby improving farming efficiency.
Implementation Method 1
A first image sensor of the image acquisition system is configured to produce color images including a plurality of color pixels
Implementation Method 2
a second image sensor of the image acquisition system is configured to produce multispectral images including a plurality of infrared pixels
Implementation Method 3
The controller generates a depth map based on bands of light common to each image in a stereo image pair
Implementation Method 4
the multispectral image for segmenting vegetation from the ground, e.g., using the Normalized Difference Vegetation Index (NDVI)
Data Source
AI summary
A farming machine identifies and treats a plant as the farming machine travels through a field. The farming machine includes a pair of image sensors for capturing images of a plant. The image sensors are different, and their output images are used to generate a depth map to improve the plant identification process. A control system identifies a plant using the depth map. The control system captures images, identifies a plant, and actuates a treatment mechanism in real time.