Abstract: To address the issues that traditional subspace-based direction of arrival（DOA）estimation algorithms require prior knowledge of the number of sources and suffer from high computational complexity，a half-spectral search DOA estimation algorithm based on the gradient jump of eigenvalues is proposed in this paper. The method first constructs a new matrix using the covariance matrix of the received signal and its complex conjugate. Simultaneously，it introduces scanning sources to form a novel covariance matrix with the scanning angle as a variable. Subsequently，eigenvalue decomposition is performed on the constructed covariance matrix. Leveraging the significant power difference between the eigenvalues of the signal subspace and noise subspace，an eigenvalue gradient discrimination function is formulated. The number of signal sources is automatically estimated by detecting the maximum points of this function. Based on the distinct demarcation characteristic between the power intensities of signal and noise eigenvalues，this method takes the gradient variation of eigenvalues as the judgment criterion and does not require prior knowledge of the number of signal sources. Subsequently，by combining the estimated number of signal sources with the low-complexity half-spectral search spectrum function，it effectively improves the DOA estimation accuracy in complex environments.