Saliency Detection Based on Fourier Single Pixel Imaging and Deep Learning

Single pixel imaging (SPI) is new type of computational imaging technology, which can obtain target object information with only one single pixel detector. It has the characteristics of low cost, wide imaging range and wide application range. The SPI using the Fourier base pattern as the illumination pattern has been proved to be a technology with both high imaging quality and high imaging efficiency. However, in complex scenes, it is still very time-consuming and resource-consuming task to obtain all the object information. Saliency detection has the property of quickly capturing regions containing important information visually from complex scenes. Therefore, we adopt the method of combining SPI with saliency detection, and propose this study of saliency detection based on Fourier single pixel imaging and deep learning models.