Abstract: Scene synthesis needs measurement for optimization. For such measurement is aimed mainly at scene optimization, and measurement results are difficult for human to interpret, people without professional interior design background need quantified evaluation to understand how a room is functionally/aesthetically/ergonomically reasonable. This paper introduces ten metrics for automatic evaluation of interior 3D scenes, comprising three aspects: functionality, aesthetics and ergonomics. First, starting from the usability and functional relationships of scene objects, scene functionality is assessed using accessibility, visibility, clearance, pairwise relationship and functional proportion. Second, considering layout composition and human perspective, scene aesthetics is evaluated by balance, alignment and harmony. Third, based on human activity simulations, ergonomic factors are analyzed by evaluating circulation and activity convenience. The proposed evaluation method can highlight the strengths and weaknesses of input scenes through individual evaluation metrics, aiding in scene optimization and design iteration. A comparison evaluation is conducted on 50 scenes from 3D-Front database, and the experimental results demonstrate that the proposed method for indoor scene evaluation performs at a level comparable to that of professional designers, with an average scoring difference ranging from 0.133 to 0.220. The validity of the evaluation results is confirmed by both professional designers and general users. In comparative experiments with PlanIT and HAISOR, the scene generation technique based on the proposed evaluation method achieves higher scores in functionality, aesthetics, ergonomics and overall evaluation, indicating a more reasonable approach to scene assessment.