At present, the phenomenon of the redseed (the maturity of cottonseed is low) exists in the cottonseed processing. Most of the domestic cottonseed processing companies are still reliant on artificial monitoring for picking the redseed. However, the visual inspection of worker has some problems such as low automation level, many subjective factors and high labor costs. Besides, the process of cottonseed delinting also results in a decreased cottonseed germination rate and other issues. To solve these problems, this paper presents a dual CCD (Charge coupled device) detection system of delinted cottonseeds color sorter which is different from the current separation technology. Through the analysis of some subsystems in the detection system, such as optical subsystem, image acquisition subsystem and image processing subsystem, and also combined with simulation analysis and tests, the effects of the proposed system were verified. Wherein the image acquisition subsystem, the cottonseed maturity information was collected with the color linear CCD, and the integrity information was collected with black and white linear CCD. The key feature information which extracted by image processing and analysis was used to make type discriminant analysis of cottonseeds. The results showed that the detection system designed optical subsystem modules could obtain the imaged with clear features. The image processing subsystem could collect material feature information which could be used in image acquisition subsystem. The designed system could ensure the realtime and stability while sorting.