Nondestructive assessment of ripeness in kiwifruit with near-infrared pulse illumination

This paper presents a nondestructive method for assessing kiwifruit ripeness using near-infrared pulse illumination at 800 nm, introducing two nonmonotonic indices (relative ripeness and the first Wasserstein distance) to quantify temporal light profile changes observed over a ten-day period.

The Gist

Imagine you have a basket of kiwifruit. You want to know which one is perfectly sweet and ready to eat, but you don't want to cut them open or squeeze them until they turn into mush. You need a way to "see" inside without touching them.

This paper is about a high-tech, non-invasive way to check the ripeness of kiwifruit using a special kind of "flashlight" and a stopwatch.

Here is the story of how they did it, explained simply:

1. The Problem: The "Steady Light" vs. The "Flash"

Usually, when scientists check fruit with light, they shine a steady beam (like a flashlight left on) and see how much light bounces back. It's like standing in a foggy room and trying to guess how thick the fog is by how much light you can see.

But the researchers in this paper thought, "What if we don't just shine a light, but we flash it?"

They used a super-fast pulse of near-infrared light (a color of light our eyes can't see, but our skin can feel as heat). Think of this not as a flashlight, but as a camera flash that goes off for a billionth of a second.

2. The Experiment: The Kiwi "Echo Chamber"

They took three Golden Kiwis (let's call them Kiwi A, B, and C) and stuck two tiny fiber-optic cables on them:

• Cable 1: The "Shooter." It fires the super-fast light pulse into the fruit.
• Cable 2: The "Listener." It waits to catch the light that bounces around inside the fruit and comes out the other side.

They did this every two days for ten days.

The Analogy: Imagine the kiwi is a crowded, foggy dance floor.

• The light pulse is a person running across the dance floor.
• The "scattering" is the person bumping into dancers (the fruit's cells), changing direction constantly.
• The "absorption" is the person getting tired and stopping (the fruit's sugars and water soaking up the energy).

Because the fruit changes as it ripens (it gets softer, sweeter, and juicier), the way the light runs through it changes.

3. The "Time-Travel" Measurement

Instead of just counting how much light came out, they measured when it came out.

• Some light zips through quickly (the "fast runners").
• Some light gets lost in the crowd and takes a long, winding path (the "slow wanderers").

By looking at the "arrival time" of the light, they could create a profile of what's happening inside the fruit.

4. The Two "Ripeness Scores"

To make sense of all this data, the scientists invented two new ways to score the fruit's ripeness.

Score #1: The "Change Meter" (Relative Ripeness)
They compared the light pattern of today's kiwi to the light pattern of the first day.

• Imagine you have a song playing. On Day 1, it's a slow ballad. On Day 5, it's a fast rock song.
• They calculated how much the "song" (the light pattern) changed from the original.
• The Result: They expected the score to go up steadily as the fruit got riper. But it didn't! It went up, then down, then up again. It was "non-monotonic" (a fancy way of saying "it didn't follow a straight line").

Score #2: The "Distance Traveler" (Wasserstein Distance)
This is a bit more mathematical, but think of it like this:

• Imagine the light pattern is a pile of sand. On Day 1, the sand is piled in one spot. On Day 5, the sand has shifted to a different spot.
• This score measures the "effort" required to move the sand from the Day 1 pile to the Day 5 pile.
• The Result: Just like the first score, this one also jumped around. It didn't just get bigger every day; it fluctuated.

5. What Did They Find?

The big surprise was that ripening isn't a straight line.

As the kiwis ripened over the ten days, the way light moved through them changed in a complex, wavy pattern.

• The Peaks: The data showed two "humps" in the light pattern. One hump represents light bouncing around (scattering), and the other represents light being soaked up (absorption).
• The Trend: As the fruit ripened, both the bouncing and the soaking up seemed to decrease. The fruit became "clearer" to the light, but not in a simple, steady way.

The Bottom Line

This paper shows that using a super-fast light pulse is a great way to check fruit without hurting it. However, it also teaches us that nature is messy. A fruit doesn't just get "riper" in a straight line; its internal chemistry shifts in waves.

By using these two new "scores," scientists can now track these complex changes, potentially helping farmers know the exact moment a kiwi is perfect to eat, without ever having to cut it open.

Technical Summary

Here is a detailed technical summary of the paper "Nondestructive Assessment of Ripeness in Kiwifruit with Near-Infrared Pulse Illumination."

1. Problem Statement

Traditional nondestructive assessment of food ripeness using Near-Infrared (NIR) light typically relies on steady-state illumination (continuous light). While effective in many contexts, the authors argue that single-frequency steady illumination often fails to capture sufficient information regarding the internal structural changes of biological tissues like kiwifruit.

The specific challenge addressed is the need for a more sensitive, nondestructive method to track the ripening process of golden kiwifruit over time. The authors hypothesize that using time-resolved measurements with pulse illumination can provide richer data on light propagation (scattering and absorption) within the fruit, allowing for better quantification of ripeness changes than conventional methods.

2. Methodology

Experimental Setup

• Subjects: Three golden kiwifruits (labeled Kiwi A, B, and C).
• Duration: Measurements were taken over a 10-day period (December 4th to December 12th, 2025).
• Instrumentation:
  • Light Source: A Time-Resolved Spectroscopy (TRS) system (Hamamatsu Photonics) emitting NIR pulses at a single wavelength of 800 nm.
  • Configuration: Two optical fibers were attached to each fruit: one for illumination and one for detection.
  • Data Acquisition: Light intensity (photon count) was recorded as a function of time ($t$) with a resolution of $\Delta t = 10$ ps over a 10 ns window. Three consecutive measurements were averaged for each day to reduce noise.

Data Processing and Indices

The authors developed a preprocessing pipeline and introduced two specific mathematical indices to quantify ripeness relative to Day 0 (the initial state):

1. Preprocessing:

   • Raw intensity data $I_j(t_k; n)$ was averaged and smoothed using a moving average filter.
   • The data was normalized to create a probability density function $p_n(t_k)$, representing the temporal distribution of detected photons on day $n$.

2. Index 1: Relative Ripeness ($r(n)$)

   • Definition: This index quantifies the change in the temporal profile of light intensity by calculating the squared intensity displacement between the current day $n$ and the baseline day 0.
   • Formula: $r(n) = \Delta t \sum d_n(t_k)^2$, where $d_n(t_k) = p_n(t_k) - p_0(t_k)$.
   • Interpretation: It represents the "area" of the difference between the current and initial light profiles.

3. Index 2: First Wasserstein Distance ($W_1(n)$)

   • Definition: Based on optimal transport theory, this index measures the "cost" of transforming the cumulative distribution of photons on day $n$ to that of day 0.
   • Formula: $W_1(n) = \int_0^T |F_n(x) - F_0(x)| dx$, where $F_n$ is the cumulative distribution function of the photon arrival times.
   • Interpretation: It provides a robust metric for the shift in the temporal distribution of light, sensitive to both scattering and absorption changes.

3. Key Contributions

• Pulse Illumination Approach: The study demonstrates the efficacy of using single-wavelength (800 nm) pulse illumination combined with time-domain analysis for fruit ripeness, moving away from multi-wavelength steady-state methods.
• Novel Quantitative Indices: The introduction of Relative Ripeness ($r(n)$) and Wasserstein Distance ($W_1(n)$) specifically tailored for time-resolved photon data. These indices allow for the mathematical quantification of subtle changes in light propagation that correlate with ripening.
• Nonmonotonic Behavior Discovery: The paper challenges the assumption that ripeness indicators change linearly or monotonically over time, revealing complex, fluctuating patterns in the data.

4. Results

• Temporal Profiles: The squared intensity displacement ($d_n^2$) plots (Figure 3) revealed two distinct peaks for each day's measurement.
  • The left peak is attributed to scattering within the fruit tissue.
  • The right peak is attributed to absorption.
  • The authors infer that both scattering and absorption properties decrease as the fruit ripens over time.
• Quantitative Findings:
  • Table 1 ($r(n)$): The relative ripeness values fluctuated significantly. For example, Kiwi B showed a peak value of 0.01069 on Day 4 (Dec 10) before dropping, while Kiwi A peaked on Day 4 as well.
  • Figure 4 ($W_1(n)$): The Wasserstein distance also exhibited nonmonotonic behavior, rising and falling over the 10-day period rather than following a steady trend.
• Conclusion on Trends: Both indices confirmed that the internal optical properties of the kiwifruit change in a complex, non-linear manner during the ripening process.

5. Significance

• Advanced Sensing: This research validates that time-domain NIR spectroscopy can detect subtle physiological changes in fruit that steady-state methods might miss.
• Single-Wavelength Efficiency: It suggests that a single, well-chosen wavelength (800 nm) combined with time-resolved analysis is sufficient for ripeness assessment, potentially simplifying sensor hardware requirements compared to multi-spectral systems.
• Theoretical Insight: The observation of nonmonotonic changes in $r(n)$ and $W_1(n)$ implies that the ripening process involves dynamic shifts in tissue density and water content that are not strictly linear, necessitating more sophisticated data analysis models for quality control in agriculture.
• Future Applications: The proposed indices could be adapted for real-time sorting systems in the food industry to determine optimal harvest or consumption times without damaging the produce.