Some challenges, such as discontinuity and insufficient robustness in hair extraction regions, unreasonable priority of traditional hair repair methods, insufficient robustness in repair sequences, and color deviation caused by overly simple sample block selection criteria are considered. The hair extraction method based on the maximum inter class variance fuzzy clustering ansd a criminisi hair repair method based on the ant colony optimization are proposed. Some important contributionss are realized, including a qualitative and quantitative proposed evaluation criterion for hair extraction repair methods. A positive balance between the repaired accuracy and the required running time is realized.