Quantification of Gaping, Bruising, and Blood Spots in Salmon Fillets Using Image Analysis

Abstract

An image analysis method was developed to quantify the gaping, bruising, and blood spots of red salmon (Oncorhynchus nerka) fillets. Images of 15 fillets with various levels of gaping were taken either with a dSLR camera, or with a video camera. Also, the same fillets were recorded using the same camera both under regular illumination, and under polarized illumination. In either case, light at an angle was used to highlight the gapes in the flesh. An image analysis method was developed that can adaptively apply an L*threshold value to the image depending on the average color of the fillet, and quantify the resulting percent of the fillet area that has an L* value less than or equal to L*threshold. Polarized lighting changed the color by eliminating artifacts resulting from reflections. It is recommended to use polarized light for this purpose. Both cameras could be used adequately to quantify defects. The proper setting of the L*threshold value depended on the camera and on the polarized light. No correlation could be found between the average L* value of the fillets and the L*threshold value. It was possible to quantify the gaping, bruising, and blood spots on the salmon fillets using this method, which can be the first step toward the automation of this operation. Gaping, bruising, and blood spots can be recognized and quantified by analyzing images of salmon fillets. Polarized light is recommended to eliminate color artifacts caused by reflected light. This can be used to automate the detection of these defects.