Image quality is not only related to distortion, but also with the quality of the luminance image. No-reference image quality evaluation has not considered the impact of the quality of the luminance image on the image's whole quality. Therefore, the luminance threshold effect is introduced to quantify the evaluation of the image's luminance image. Based on the BRISQUE algorithm and with fast fading distortion as an example, experiments were carried out for the 50 images acquired by adjusting their background luminance. The distortion image was stratified into the incident and reflected components. Luminance threshold algorithm was adopted for the incident component (luminance image) and BRISQUE algorithm for the reflection component (reflect image). A new method for evaluating the quality of no reference image was put forward. The Pearson correlation coefficient (PCC) of the proposed algorithm was 0.9982, and the Spearman's Rank ordered Correlation Coefficient (SROCC) value was 0.9741. According to the experimental data, this algorithm has better consistency with the subjective evaluation of human vision than the BRISQUE algorithm and fits the human visual perception.