The invention discloses a digital circuit board self-diagnosis system based on the feature compression technique. The digital circuit board self-diagnosis system based on the feature compression technique comprises a test vector generating module, a space compression module, a time compression module and a diagnosis module, wherein test vectors are added to N modules to be tested of a circuit board respectively through the test vector generating module, so that test response K*K-order square matrix data are obtained; the test response K*K-order square matrix data are added according to phase modes through the space compression module, so that 2K odd-even check values are obtained; time compression is conducted on the 2K odd-even check values through the time compression module so that a line detection value and a row detection value are obtained; the line detection value and the row detection value are compared with a prestored standard line detection value and a prestored standard row detection value respectively through the diagnosis module, and the position information of a corresponding faulted module is displayed according to the line detection value and the row detection value when the line detection value and the row detection value are not equal to the prestored standard line detection value and the prestored standard row detection value respectively. The invention further provides a digital circuit board self-diagnosis method based on the feature compression technique.