Role of Color and False Color in Object Recognition With Degraded and Non-Degraded Images

摘要

Recent technological advances in the design and manufacturing of night vision multispectral sensors now allow spatially registered imagery provided by each of the sensors to be combined within a single fused image for display to an end user. The product is a multispectral false colored rendering of the imaged scene. The use of false color in fused imagery may facilitate object recognition, providing contour information of the objects present in the scene, but incongruently colored fused imagery, may be disruptive of perceptual performance. This study investigated if the use of false color imagery compared to natural color, imagery was helpful or not in object recognition. Subjects' reaction times (RTs) and error rates were measured in a standard naming task. Stimuli consisted of photographs of food objects that had been manipulated in color (natural color, false color, natural grayscale, and false grayscale) and noise (three levels). The results of the experiment showed similar differences in RTs between color images (natural or false) and their grayscale counterparts at different levels of noise, indicating that both color conditions were similarly helpful in object recognition. These results give an indication that false color may be useful in multispectral sensors based on its facilitation of image segmentation with shape degraded images.