Hyperspectral Imaging Device Accurately Detects Changes in Plant Health

Jian Jin, a professor at Purdue University, has developed a novel, potable sensor that gives farmers and plant scientists a more accurate way of determining the crops’ health while collecting up-to-date data that will prove valuable to federal/state officials and others.

An assistant professor in Purdue’s Department of Agricultural and Biological Engineering, Jian Jin believes that his new hyperspectral imaging device will be extensively used by farmers and plant scientists at the national and international levels. The hyperspectral imaging device has the ability to scan a plant for physiological traits, for example, chlorophyll, moisture, and nutrient levels, and also different types of disease symptoms and chemical spraying effects to establish whether it is under stress or healthy.

Jin informed that his newly developed hyperspectral imaging device will assist farmers to identify changes in plant health in the field in a matter of hours to days much before they can be seen by the naked eye. The device will also enable farmers to make the required modifications to grow more amounts of food with the help of fewer resources, for example, by reducing the use of water and fertilizers.

“My vision is this sensor will allow household farmers walking through a field to use a handheld device and a smartphone to get the same information available from very expensive phenotyping systems constructed by big companies and big universities in recent years,” said Jin. “We have 600 million farmers worldwide, and very few of them are benefiting from high-end plant sensor technologies. Now, with this handheld device, most farmers can benefit.”

This novel technology complements Purdue’s “Giant Leaps” commemorating the university’s worldwide advancements made in artificial intelligence, space, health, and sustainability highlights as part of the university’s 150^th anniversary. Aligning to the yearlong celebration’s Ideas Festival, those four themes are designed to demonstrate Purdue as an intellectual hub solving real-world problems.

The sensor, which is capable of scanning a plant in just a matter of five seconds, can detect scores of color bands in every pixel as opposed to the three bands of color perceived by conventional cameras. There is another version that even shoots a burst of fluorescent light off the plant. Both sensors are used for measuring the plant’s nutrition and stress levels.

“We implemented both the hardware and software technologies into a handheld device that is light and easy to carry,” said Jin.

The sensor combined the plant features prediction models created by Purdue researchers and the sophisticated image processing algorithm. The models were designed with the help of Purdue’s database that contains years of plant research assays in the field as well as the greenhouse. Moreover, the models are continuously enhanced and updated.

“So we always have the most accurate predictions for the farmer,” said Jin.

Over the past 10 years, there has been a rapid advancement of plant phenotyping as technology is progressively being used to enhance efficiency on the basis of present conditions rather than farmers depending on historical data and regional conditions to make sound decisions. In the majority of farms, plant health is manually, which lacks efficiency and precision.

Today’s devices used by plant scientists clamp down on a leaf and determine the health of just a part of the plant. Jin claimed that his sensor is more accurate than such devices.

“Due to multiple technical reasons, the sensor’s prediction quality is much more accurate than any other types of crop imaging sensors that people have in the existing market,” Jin stated. “It’s also constantly getting better because we scan plants every day and are upgrading both hardware and software technologies.”

While the sensor is self-contained, option is available to users to upload the measurements with geographical locations to a web-based cloud map service devised by Carol Song and her group at Purdue’s Advanced Computing Group. Based on the sensor measurements, the system produces the plant’s nutrition and stress heat maps and gives interactive ag data questioning functions at both regional and farm levels.

This advanced digital ag map system integrated with sensor data has the potential to support a number of promising applications. For instance, the data gathered will give useful information to federal and state officials regarding steps they can adopt to assist farmers during the periods of severe crop stress and also information about the types of crop yields that can be anticipated.

“If we can successfully distribute the sensors around the region, we can generate this digital ag map service to monitor the plant growth all over the region—which areas are under stress and which areas are having a good performance,” he stated.

Jin’s team at ABE is working on this device’s automation. Last winter, together with his graduate students, Jin worked with a senior design team from Purdue’s School of Mechanical Engineering and effectively used a robot to check the leaves with the sensor automatically in the greenhouse. Using machine vision, the robot detected the target leaves and carried the sensor over there for a rapid scan operation along the natural slope of the leaf. Motivated by the success in the greenhouse, Jin and his research team are working on the design of a next robot in the farm field environment.

This unique robot system may appear like a spider transformer, traveling between rows of crops with each leg fitted with a sensor, scanning and waving the leaves in the field with an extreme speed. This prototype is expected to function during the 2019 growing season.

Jin is seeking collaborators who can lead in commercializing the novel device, particularly in mass manufacturing and marketing. According to him, the best approach would be to make the devices cost-effective with the data being where the value is.

“We hope to get a lot more data so we can have more valuable data services,” he stated. “We have great team work at Purdue to make it happen. Besides the engineers from ABE, the sensor’s development has been greatly supported by breeders and biologists at Purdue, including professors Mitch Tuinstra, scientific director of the Purdue’s Institute for Plant Sciences, and Tony Vyn, the Henry A. Wallace Chair in Crop Sciences. Carol Song and her team of data scientists from Advanced Computing Group provided the GIS map functions. Gerald Shively from Agricultural Economics has been promoting the application of the device as a social scientist.”