Automatic Visual Inspection of Granular Biomaterial Under Active Illumination for Accurate Detection of Micro Polluants

In this paper, to improve the process of quality control in the production line of the biomaterial Nukbone®, we present an Automatic Visual Inspection system with active illumination and an iteractive interface that accurately detects micro pollutants beyond the limits of the human sight and simplify their identification and location for remotion. The system, inspired in the current inspection process that manually detects and remove pollutants off, introduces an illumination strategy to opticaly highlight foreign particles in the sample by integrating numerically filtered images of multiple sources of ligh for enhanced discrimintion properties. It improves real time detection of tiny particles, by contouring them using the Suzuki and Abe method, and displaying the results in a visual interface that shows the inspected sample in a sensible image where, areas with a positive detection are indicated can be selected, to observe, in a magnified image, the detected particles contoured and superimposed on the real image of the biomaterial. We describe the system and its functionality, and illustrate its performance with some examples under real conditions that shows its capacity detecting imperceptible particles in a manual inspection and reducing significatively the inspection time.