Image reconstruction using single-pixel color ghost imaging

Abstract

A simple method is proposed to recover color image by using single-pixel ghost imaging (GI) technique. A color image (RGB format) is firstly converted to its indexed image format before it is measured, and conventional GI technique is used to measure and recover this indexed image. Quality of the recovered image using conventional correlation algorithm of GI is low even using a large amount of measurements. It is difficult to convert the low-quality indexed image back to the RGB image. To reconstruct color image with high quality, a Gerchberg-Saxton-like algorithm is used to further improve contrast of the recovered indexed image. This algorithm can lead to a nonlinear growth of signal-to-noise ratio (SNR) value with respect to the number of measurements. Hence, after a dramatic improvement, a large SNR value of the indexed image can be achieved and a comparatively high-quality color image can be obtained from the improved indexed image. Compared to traditional method in GI which treats a color image as three independent channels (i.e., red, green and blue), the proposed method is simpler, because it reduces the complexity by converting original three channels into a single channel. Besides, since the recovered indexed image is processed by a Gerchberg-Saxton-like algorithm, high-quality color images can be obtained. It is worth noting that the proposed method uses a post-processing strategy, and it does not add the complexity. Some computational results are presented to prove that the proposed method is feasible and effective.