Images are the best source of information for human beings and help deeply understand every type of surroundings. Therefore, computer science has tried to bring the quality of visual information into its activities and get the benefits out of it.
In computer science, image processing consists of the manipulation of images using computers. Therefore, images are seen as a source of data and information, that can be read if the image is analyzed in a proper way.
In this article, we will walk through a short description of how image processing works and how it can be employed in precision agriculture.
How image processing works
To be processed, an image is converted into a number matrix, which can be easily read by a computer.
There are many types of processing and allow for different activities, like image enhancement, image restoration, image compression, or image analysis. The latter turns out to be particularly interesting since it permits to extract specific information directly from an image. The analysis can be run by looking at the edges of images (i.e. image extraction), at their colors (i.e. texture analysis) and the movements detected going from an image to another (i.e. motion analysis).
The process follows some general basic steps. The first one is image acquisition, by which the material to be analyzed is collected. Then, in preprocessing the image is optimized, to get the best out of it. In the segmentation step, the image is segmented into smaller parts, that can be then more easily compared and analyzed by the computer. The data is then made suitable for computers and is made recognizable thanks to some labels that are assigned to different pieces of data. These pieces can be now put together, to be interpreted, and to get valuable information out of them.
How it started
The first studies on image processing date back to the 1960s, when it was used to collect data on the moon and the sun during the lunar walk. Since the technology was quite expensive, image processing has been used mainly in the scientific field, until the quick development of computer starting from the late 1990s.
Today, image processing has a wide variety of applications. Besides all those techniques to edit pictures, the data collected through images are used also in the medical sector, in scientific research (e.g. spatial researches), in fingerprints/iris/face recognition technologies, in remote sensing (e.g. satellite images) and industrial applications (e.g. quality check, sorting, etc.).
Important applications of image processing in agriculture
Image processing is extensively used in agriculture as well. The main advantage of the technology is that it is nondestructive, meaning that it can provide insightful information about crops without even touching them.
The main applications are related to four categories. The first one is crop management, especially regarding pest and disease detection and irrigation methods. The second one is related to the analysis of crop leaves and skin conditions, which helps with the identification of nutrient deficiencies and plant content. Also, fruit quality inspections and sorting are possible, by those images and suggestions collected thanks to advanced machine learning techniques. Last but not the least, in crop and land estimation, image processing is used for the Geographic Information System and color & texture segmentation of crops.
Want to learn more about precision agriculture and its applications? Check out our blog post here.
What does Pixofarm do with image processing?
Pixofarm is a solution based on image processing, where the source of data is a number of pictures of fruits taken directly in the orchard throughout the entire season.
Aimed to provide data about the growth of fruits and predicted information on the outcome of the season, Pixofarm brings this technology to an area at the beginning of its development, meaning yield monitoring. While more developed solutions are about pest detection, quality inspection, and crop management mainly for extensive crops, Pixofarm is focused on horticulture.
The technology is trained to precisely measure fruit sizes just from a bunch of pictures taken with almost any phone since it is dynamically adaptable based on the phone hardware characteristics (especially the camera and lens specifications).
The application is designed to be as inexpensive as possible from the computational viewpoint, storage space, bandwidth, and power consumption. Plus, it gives access to many users on the same account.
The main features are:
Object detection machine vision: Pixofarm can recognize the apple and focus on its shape.
Fast and accurate segmentation using watershed, and computational geometry algorithms.
Leaf tolerance and detection stability in outdoor noisy conditions using adaptive filters.
Color independent image processing using algorithms and geometry tools.
Stereo vision and point cloud analysis for 3D reconstruction and better sizing.
Image processing is a powerful tool that can enable precise and insightful analysis
In conclusion, image processing is a powerful tool that can enable precise and insightful analysis in many different fields. Thanks to the ease of acquiring images today, this technology is gaining more and more importance, even in the agriculture sector.
In a context of strong image processing growth, Pixofarm wants to bring state-of-the-art technologies into orchards and to develop a lighter and more advanced image processing, to move a further step toward an enhanced Farming 4.0.
Curious to know more about Pixofarm? Subscribe to our newsletter below or book a demo here to discover more about our solution.
· Eduardo A.B. da Silva, Gelson V. Mendonça, in The Electrical Engineering Handbook, 2005
· E. Suganya, ... M. Karthiga, in Internet of Things in Biomedical Engineering, 2019
· image processing applications https://medium.com/@danilloleite82/the-history-of-image-processing-information-technology-essay-a30690c352f4
· Image processing and related fields http://fourier.eng.hmc.edu/e161/lectures/e161ch1.pdf
· Pandurgn J.A., Lomte S.S., Digital image Processing application in agriculture: a survey, 2015