Recycling scrap metal is an important way to protect the ecological environment. Design effective yet efficient techniques to automatically identify recyclable scrap metals is an important task within this topic. Due to the advantages of fast response and high accuracy, laser-induced breakdown spectroscopy (LIBS) recently played an important role in the mineral identification. However, the identification accuracy of peak-seeking is greatly affected by the data quality of the LIBS spectrum, whereas machine learning methods may be greatly affected by the number of training data. By considering the above open issues, this paper proposes a hybrid algorithm based on support vector machine (SVM) and element peak-seeking. By investing the identified difference of the major element (with the largest composition in the alloy) and the general element (with composition more than 1% in the alloy) between peak-seeking and SVM, three integration types (i.e., rejection, partial acceptance, complete acceptance) are defined. The final recognition result is generated according to different integration types and the corresponding integration methods. To verify the feasibility of the proposed approach, a simulated alloy LIBS database was established based on 31 metal elements and the simulated alloy LIBS data according to their compositions. Comparing with the result obtained by only using SVM, the proposed method greatly improved the recognition accuracy. The accuracy of identifying all general elements increased from 8% to 74.5%. Experimental results confirmed the effectiveness of the proposed method in identification of general metal elements in terms of higher detection accuracy.