The invention provides a meat product freshness detection method based on multisource perceptual information fusion. According to the method, a near infrared spectrometer, a computer vision system and an electronic nose are used for acquiring internal and external information of a sample simultaneously and respectively, the feature extraction and fusion treatment are performed on the acquired information, and the freshness grade of a meat product is evaluated according to a freshness grade evaluation standard of the meat product. The method comprises the steps of establishing the freshness grade evaluation standard of the meat product, establishing a freshness grade forecast model of the meat product, and performing freshness grade evaluation on the sample to be detected. The method comprehensively uses the spectrum, image and smell fingerprint information, comprehensively determines the internal and external quality of the meat product, and can realize quick, simple, convenient and objective grade evaluation of the freshness of the meat product.