The invention discloses a rapid primary separation system and method for naturally curled tobacco leaves after baking, and the system employs a camera as image collection equipment, employs a composite algorithm combining an OTSU algorithm and an edge recognition algorithm as a recognition algorithm, employs a composite parameter formed by combining an RGB parameter, an HSV parameter, a leaf profile parameter and a leaf surface texture parameter as a grading parameter. A Bayesian classifier is adopted as a grading algorithm, automatic intelligent analysis and processing of a series of steps of image collection, target recognition, parameter yeast extraction and intelligent judgment of the naturally curled tobacco leaves after baking are achieved through system curing, the grading accuracy is improved, pre-checking and packaging can be achieved at home of tobacco growers, direct grading is achieved in a purchasing site, the working efficiency is greatly improved, the method solves the problems of high cost, low efficiency and non-standard grade qualification rate in the primary separation link of tobacco growers, and plays an important role in improving the grade purity of tobacco purchasing, promoting high-quality development of tobacco production and keeping continuous, stable and healthy development of tobacco production.