With the continuous improvement of people’s living standards, consumer demand for apple has shifted from quantity to quality. However, for small-scale farmers, the detection equipment developed on the market based on spectrometers is expensive, resulting in difficulty in purchasing, and thus it is impossible to achieve high-quality fruit screening. Therefore, the object was to develop a lowcost, portable apple multiquality parameter integrated detection device. Firstly, diffuse spectral information of 240 Fuji apples was acquired based on the visible/near-infrared spectroscopy platform. After SG smoothing and multiple scattering correction algorithm pretreatment, the characteristic wavelength of soluble apple solid, titrable acid and pulp firmness was extracted by random frog algorithm. So the ten shared characteristic wavelengths of the three parameters, namely 420nm, 480nm, 550nm, 580nm, 640nm, 680nm, 705nm, 940nm, 980nm and 1044nm, were selected. On this basis, a detection method combining a characteristic narrow-band LED light source and a photodiode was proposed. Then, hardware systems such as diffuse reflection detection optical path, narrow-band LED ring light source, detection probe, control circuit and so on were designed. Secondly, the diffuse reflection characteristic voltage intensity of 144 apple samples was obtained by the detection device. Then the multivariate linear regression models of soluble solids content, titratable acid and firmness were respectively established by the obtained information. The correlation coefficients of the prediction set were 0.8129, 0.8073 and 0.7736, and the root mean square errors were 0.6036°Brix, 0.0636% and 1.7325N, respectively. Based on the QT development tool, the realtime control and analysis software of the device was developed in Python language, and the multiquality parameter prediction model of apple was implanted to realize the simultaneous detection and analysis. Finally, to test the precision and stability of the device, another 46 samples were selected and each sample was tested eight times. The predicted results of the device were acceptable, and the correlation coefficients of soluble solids, titratable acid and pulp hardness of apple were 0.8096, 0.7962 and 0.7589, and root mean square errors were 0.6973°Brix, 0.0703% and 1.8323N, respectively. The maximum variation coefficients of the resampling were 0.0106, 0.0116 and 0.0062, respectively. The results showed that the low-cost, portable nondestructive testing device based on multi-featured narrow-band LED light source can realize the simultaneous detection of multiple quality parameters of apples, which can meet the needs of farmers for field production and e-commerce sales of high-quality fruit screening.