With the attention of the country to the ecological environment, nuclear energy has been vigorously developed, and the deep internal thread on the nuclear reactor dome cover is easy to appear scratches and collapse problems during operation. In order to solve the defect identification problem of deep hole internal thread, this paper puts forward a system to identify the defect of deep hole internal thread by using rotating linear array CCD. The system includes the image acquisition module of internal thread surface, the core control module of motion mechanism and the defect identification module. The experimental prototype of the detection system is used to complete the drive control of the rotating linear array CCD, and the defect detection system of the deep inner thread is verified by using multiple thread images, which meets the expected requirements. The acquisition method of the surface image of the deep internal thread proposed in this paper has a certain reference significance for the acquisition of two-dimensional planar expansion of the internal thread. The designed defect detection algorithm has a certain reference value for recognizing the defect images containing texture features.