Detection of bone fragments in poultry meat remains a challenge. In the industrial deboning process, bone
fragments often remain in chicken fillets, escaping manual or x-ray machine detection. A new method was proposed to
compensate for x-ray absorption variations, which occur due to the uneven thickness of fillets. Imaging detection
algorithms were developed to integrate thickness and x-ray images and to produce a thickness-compensated x-ray image
for the extraction of bone fragment signals. Plastic molds with four different slopes were designed for forming poultry
meat for conceptual tests and imaging algorithm development. Experimental results from sample analyses demonstrated
that the four types of frequent and hard-to-detect bone fragments could be detected no matter where they were located in
chicken meat of uneven thickness. The proposed imaging method eliminated the false patterns and enhanced the
sensitivity of x-ray in bone fragment detection. The technique has the potential for non-destructive internal detection of
hazardous materials in food products with uneven thickness.