.png){kind=icon}
.png){kind=icon}
Return
Ghost imaging target classification through deep sequential feature extraction
.png){kind=icon}
PRE
AI
DOI：10.1016/j.optlastec.2025.113556.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• An LSTM-based approach for image-free object classification in CGI enables target detection directly from bucket signals, treating the GI process as a temporal signal classification task instead of an imaging problem. • The network leverages the LSTM’s gated mechanism for noise suppression, long-term dependency capture, and sequential feature extraction, achieving classification even at low sampling rates (down to 1 %). • The method relies on minimal hardware requirements, utilizing a bucket detector with no spatial resolution for classification, making it applicable to fields with hardware constraints.
Keywords:
LSTM
object classification
CGI
bucket signals
noise suppression
Journal
O
IF:
5
Papers:
1.8K
Citations:
3.5W