Development and application of "The Two Image" method for accurate object recognition and color analysis

Segmentation (separation of object from background) is a key operation in image analysis and machine vision methods. When objects contain colors similar to the background, segmentation becomes difficult. The "two image" method overcomes this difficulty by defining the object from the silhouette obtained from a backlighted image (image I), and performs color analysis using a front lighted image (image II) using segmentation from image I. Segmentation of a silhouette is much easier than a color image. Since there is no need to have different colors for the object and the background, the color-biasing of semi-transparent objects is eliminated. Salmon pieces, squid tubes and peeled raw shrimp were given as examples of application of this method, with special emphasis on the analysis of semi-transparent objects (shrimp). (c) 2012 Elsevier Ltd. All rights reserved.