Soil is the basis of agricultural production. The determination of soil nutrient content is determined by the determination of soil nutrient content, and it has a certain influence on the growth of crops. Therefore, the detection of soil component content has gradually become a research hotspot at home and abroad. Multi-spectral technology utilizes the difference in the physical structure and chemical composition of an object. Under the same conditions, the object is irradiated with different light reflections to obtain different reflectances on the corresponding spectral bands, and then the acquired spectral data is analyzed to identify the target. In recent years, the application of multi-spectral technology has provided a new idea for the detection of soil component content which is helpful for the accurate detection of soil component content, and contributes to the realization of non-destructive real-time online detection technology and precision agriculture. In this paper, the related pieces of literature on the application of multi-spectral techniques in the soil composition index of soil water, organic matter, NPK, heavy metals and soil salinity in the past six years are reviewed. The characteristics of multi-spectral imaging technology are analyzed, and the multi-spectral is briefly described. The detection process of soil composition content by technology focuses on the research progress of multi-spectral technology in soil component content detection, and prospects the future development trend of multi-spectral technology in soil component content detection, and proposes the future technology development direction: machine learning. The unsupervised and supervised model of the algorithm can analyze the data in different actual measurement environments, reduce the influence of spectral data with uneven distribution of soil components on the modeling results; multi-spectral images combined with panchromatic images to obtain multi-spectral panchromatic bands, In the multi-spectral soil component content detection, the accuracy and accuracy of the prediction model are improved. In the multi-spectral data preprocessing process, two or more algorithms are combined to process the spectral data more effectively.