Online Detection of Chicken Carcass Contaminants via Machine Vision

This paper describes a technology for detecting contaminants on the surface of chicken carcasses using machine vision. This technology involves capturing images of chicken carcasses at specific wavelengths (500 nm and 710 nm) using an industrial camera system equipped with filters. The images are then processed to enhance quality and define contamination regions through various image processing techniques like median filtering, gray scale enhancement, and binary thresholding. The goal is to identify and treat contaminated areas on the chicken carcasses automatically.

For non-experts, imagine a camera system set up along a chicken processing line, taking pictures in special light that highlights contaminants like feces, blood, or bile that might not be visible to the naked eye. This system processes these images quickly to spot any dirt or unwanted substances and then activates a spray to clean the contaminated areas right away. This technology helps ensure the chicken that reaches consumers is clean and safe, maintaining high quality and reducing health risks.

Filter Selection

n the described system, two specific filters are used: one for the wavelength of 500 nm and another for 710 nm. These filters are crucial because they enable the camera to capture images that emphasize the differences between the chicken carcass contaminants (like cecal feces, blood, and bile) and the chicken skin. By filtering the light to these specific wavelengths, the system enhances the visibility of the contaminants against the background of the chicken carcass, making it easier for the image processing algorithms to detect and segment these areas.

The filters play a significant role in:

1. Isolating Key Wavelengths: These filters allow only light of specific wavelengths (500 nm and 710 nm) to pass through. These wavelengths were chosen based on their ability to distinguish contaminants from the chicken skin more clearly than other wavelengths.
2. Enhancing Contrast: By limiting the light to these wavelengths, the filters help increase the contrast between the contaminants and the chicken skin in the captured images. This contrast is crucial for effective image analysis, as it simplifies the process of detecting and distinguishing between clean areas and areas with contaminants.

If these filters were not used, the camera would capture images with a broader range of wavelengths, potentially including many that do not significantly differentiate contaminants from the chicken carcass. This could lead to:

• Reduced Contrast: Without the filters, the contrast between the contaminants and the chicken skin could be much lower, making it difficult to distinguish and accurately identify contaminated areas.
• Increased Noise and False Detections: Capturing unnecessary wavelengths might introduce noise into the image, leading to inaccurate detections and higher rates of false positives or false negatives in identifying contaminants.
• Inefficiency: Overall system efficiency would decrease as the image processing algorithms might struggle to process the additional irrelevant data, possibly requiring more complex and computationally intensive methods to achieve the same level of accuracy.