Design and Implementation of Artificial Intelligence Image Detection System Based on Plane Features

Artificial intelligence has entered modern people's life, and image detection is the cornerstone in many application fields, especially in the field of Internet of Things. Image is a special data format. If we want to obtain different connotation information according to different images, we need some image detection technology, and also need some image feature algorithms. In this paper, we propose to realize registration by using plane features, which are obtained by segmenting depth images. At the same time, a new geometric feature invariant-feature ratio is proposed, and a new affine invariant image feature descriptor is constructed based on it. By constructing a series of straight lines intersecting the target image, the descriptor represents the image with the positional relationship of a series of collinear points. Experimental results show the accuracy and robustness of this registration method.