WiMi Hologram Cloud Inc. (NASDAQ: WIMI) (“WiMi” or the “Company”), a leading global Hologram Augmented Reality (“AR”) Technology provider, today announced that its R&D team is developing an optical scanning holography reconstruction algorithm based on convolutional neural networks (CNNs), which process interferometric images by CNNs to generate high-quality holographic images. Features in the interferometric images are extracted using convolutional kernels, which are then used to create holographic images. Compared with traditional reconstruction algorithms, using CNNs can reduce noise and artifacts and improve the resolution and sharpness of the reconstructed images. In addition, CNNs can accelerate the reconstruction process through parallel processing and optimization algorithms. High-quality holographic reconstruction can be achieved while reducing computational complexity and data requirements.
WiMi’s algorithm is end-to-end learning and reconstruction of optical scanning hologram images using a CNN model. Specifically, the algorithm first takes the acquired optical scanning hologram image as input and feeds it into the CNN model. Then, the CNN model automatically extracts high-level features from the input hologram image and continuously adjusts the network parameters to minimize the reconstruction error through a back-propagation algorithm. Eventually, the algorithm outputs a high-quality holography reconstruction result. The technical process includes steps of data acquisition, data pre-processing, CNN model training, model testing, result evaluation, and optimization and improvement.
WiMi’s algorithm can automatically extract features, learn more complex and advanced features from the data, and obtain higher-quality reconstruction results than traditional holographic image processing algorithms. And the whole reconstruction process of this algorithm is carried out in an end-to-end framework, simplifying the process and improving holographic reconstruction efficiency.
Optical scanning holography reconstruction algorithm based on convolutional neural network is an up-and-coming technique. It has significant application value and has a broad application prospect in digital holographic imaging, medical image analysis, 3D object recognition, and other fields. With the continuous development and improvement of technology, it will be applied in more areas and bring more convenient and efficient services to people.
In the future, WiMi will also combine the algorithm with other computer vision and image processing technologies to achieve more comprehensive and accurate data analysis and image processing. In addition, WiMi will also consider applying the technology to practical production, such as industrial inspection and non-destructive testing. WiMi hopes to achieve more intelligent and efficient production management in the coming years.
