A least square based minimum error thickness compensation approach for physical contamination detection is presented in this paper. For conventional X-ray screening of poultry fillets, detection errors are significant due to the unevenness of poultrymeat, causing difficulties in detecting small and low-contrasted foreign objects in poultry meat. Laser range image is used to compensate the thickness variance effect. The algorithms and experimental results for image registration and compensation arepresented.
展开▼