digital image analysis aided Adulteration Detection Data Set for Fennel Powder adulterated with wood

Adulteration Detection of Powder with Wood
About Fennel Powder
Contains dietary fiber, vitamins (A, C, B-complex),
Adulteration Issue
Presence of rice husk reduces purity, nutritional value, and may cause digestive discomfort.
Visual detection is difficult as powdered rice husk blends with Fennel powder.
Detection Methodology
Dataset creation: Images of pure and adulterated Fennel powder with varying rice husk percentages, labelled by experts.
Image Preprocessing: Techniques like grayscale, RGB, HSV, median smoothing, adaptive thresholding, and edge detection to enhance visual differences.
Feature Extraction: Identify texture, color, and surface pattern differences between pure and adulterated samples.
Sample preparation
Full-grown Fennel powder of local high-yield variety were collected from [maliha, Malda, west bengal, india]. Time for conducting 9.00 am to 5.00 pm. The final moisture content of the dried seeds was 4–5% (dry weight basis), and The total ash content of Fennel powder typically ranges from 5% to 8%, depending on the source and processing methods
The dried seeds were ground using a mixer grinder (500 W, grinding jar capacity 0.5 L, hybrid motor) for 9 min (3 min per cycle, total 3 cycles, 18,000 rpm). The resulting powder was passed through a 60-mesh sieve, and the undersized fraction was considered pure Fennel powder.
Rice husk was collected from a local rice mill, cleaned to remove dust and impurities, and oven-dried at 50 °C for 4 h to ensure complete dryness. The dried husk was ground using a hammer mill and passed through a 60-mesh sieve. The undersized fraction was considered as the adulterant.
Fennel powder was mixed thoroughly with rice husk at different proportions (0%,10%, 20%, 30%, 40%, and 50%, 100% by weight) to prepare a homogeneous mixture of adulterated samples.
Image Acquisition
In total, 7 sets of samples were prepared: 100% pure Fennel powder, 100% rice husk powder, and 10 sets of adulterated samples containing 5–50% rice husk by weight. A total of 20 equal-sized cells were drawn on a white A4 paper sheet (210 mm × 297 mm, 80 gsm). The samples were spread over the paper in a thin layer. Thus, 20 equal-sized cells were loaded with similar samples (Fig. 1). In total, 7× 20 = 140 images were captured.
Each of the cells was photographed separately using a Realme C63 5g smartphone (Android 15 Fun touch unisoc T6 12, processor, MediaTek Dimensity 6300MP f/2.2 (wide), 2 MP f/2.4 (macro), 8 MP f/2.2 (front), 1/3.06", 1.12 µm) without flash. The perpendicular distance between the sample surface and the camera lens was maintained at 32 cm.