The invention discloses a multi-feature information fusion based insulator state detection and diagnosis method. Through introducing a computer vision analysis technology with artificial intelligence significance, an existing inspection system is enabled to have a recognition ability of the human eyes and an analysis ability of the human brain. Potential threat events such that a power transmission line insulator is damaged are recognized and analyzed through visual analysis, real-time early warning is achieved, and the accident rate is reduced.