Crop Disease Detection in Large Field of View Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional image processing methods fail to accurately detect crop disease symptoms from large field of view images, as disease symptoms on individual crops can be minor features and often go undetected.
Innovation Solution
A computer system is configured to build models for identifying candidate regions capturing single leaves and detecting disease symptoms, using a combination of histogram of oriented gradients (HOG) and convolutional neural networks (CNNs) to analyze large field of view images, applying efficient and robust computational algorithms in two stages to focus on potential regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image processing methods are used on large field of view images, then processing speed is maintained, but detection accuracy of disease symptoms deteriorates because symptoms on individual crops are minor features that go undetected
Solution Approach 1:
The patent divides the large field of view image into multiple smaller sub-images or regions of interest. This segmentation allows the system to focus computational resources on detecting disease symptoms in each smaller region, improving detection accuracy while managing system complexity through distributed processing.
Solution Approach 2:
The patent transforms the image processing problem by creating an image pyramid with multiple scales or resolutions. By analyzing the image at different dimensional scales, the system can detect disease symptoms that appear as minor features in the original large image but become more prominent when viewed at appropriate scales.
2Measurement precision
If traditional image processing methods are used on large field of view images, then computational resource usage is limited, but detection capability deteriorates because disease symptoms constitute minor features that are easily missed
Solution Approach 1:
By segmenting the large image into smaller regions, the patent reduces the computational power required for each individual processing task. The system can apply detection algorithms to multiple smaller regions in parallel or sequentially, achieving better overall detection accuracy without requiring excessive computational resources for a single full-image analysis.
Solution Approach 2:
The patent applies partial action by focusing computational resources only on regions that are likely to contain disease symptoms, rather than processing the entire large image uniformly. This selective processing improves detection accuracy while conserving computational power by avoiding unnecessary analysis of regions without symptoms.
3Measurement precision
If comprehensive analysis of all regions in large field of view images is performed, then detection accuracy improves, but processing time increases significantly
Solution Approach 1:
The patent segments the image processing task into smaller, independent regions that can be analyzed in parallel or in a prioritized sequence. This reduces the total processing time while maintaining detection accuracy, as the system doesn't need to perform exhaustive analysis of every pixel in the entire large image simultaneously.
Solution Approach 2:
The patent performs preliminary actions by first identifying regions of interest or potential disease locations before conducting detailed analysis. This preliminary filtering step reduces the amount of time required for comprehensive analysis by focusing subsequent processing only on areas that are likely to contain disease symptoms.
Data Source
AI summary
In an embodiment, a computer-implemented method of detecting infected objects from large field-of-view images is disclosed. The method comprises receiving, by a processor, a digital image capturing multiple objects; generating, by the processor, a plurality of scaled images from the digital image respectfully corresponding to a plurality of scales; and computing a group of feature matrices for the digital image. The method further comprises, for each of the plurality of scaled images. selecting a list of candidate regions from the scaled image each likely to capture a single object; and for each of the list of candidate regions, performing the following steps: mapping the candidate region back to the digital image to obtain a mapped region; identifying a corresponding portion from each of the group of feature matrices based on the mapping; and determining whether the candidate region is likely to capture the single object infected with a disease based on the group of corresponding portions. In addition, the method comprises choosing a group of final regions from the lists of mapped regions based on the determining; and causing display of information regarding the group of final regions.