Abstract: [Objective] This paper proposes a method and technology for quantitative evaluation of the appearance quality of sawn timber, and explores the feasibility of digital detection and evaluation of the appearance quality of sawn timber in order to provide basic theoretical and technical supports for real-time online detection of appearance quality.[Method] An on-line detection system for surface defects of sawn timber, which consists a sealed darkroom, a light source, an industrial camera, etc., was established, and color images of the sample sawn timbers were collected under a stable light environment. Based on the image processing technology, the software experiment system was developed, which realized the detection and recognition of sawn timber defects. The detection and recognition rates of knot, hole and crack defects were 81.7%, 91.2% and 72.6%, respectively. The appearance quality evaluation model of knot, hole and crack was also established respectively. Based on these, a comprehensive quantitative evaluation method of appearance quality was proposed. The scientific value and feasibility of the method was verified by comparing experiments with national standards.[Result] By comparing the quantitative evaluation method with the grade index of sawn timber material, the linear correlation coefficient of the two is 0.85. The distribution of the comprehensive quantitative evaluation values corresponding to the grades of the sawn timber material grades 2 and 3 is relatively scattered, and the data between the grades is partially crossed. Besides, there is almost no crossover of the corresponding comprehensive quantitative evaluation value distribution between the other levels. By comparing with the grade index of sawn timber material, the linear correlation coefficient between the two is 0.88. The distribution of the comprehensive quantitative evaluation values corresponding to the grade 4 of the sawn timber material index level is relatively concentrated, and there is no data intersection with other grades. The timber material index grades are grade 1, grade 2, and grade 3. The distribution of the comprehensive quantitative evaluation values corresponding to the adjacent grades is relatively scattered, and the data between the grades is partially crossed. There is no intersection of data distribution between non-adjacent levels.[Conclusion] The research shows that the linear correlation between the comprehensive quantitative evaluation value and the grade of sawn timber material index determined, according to the relevant national standards, is relatively good. The linear correlation between the comprehensive quantitative evaluation value and the appearance quality requirements of the integrated timber board determined, according to the relevant national standards, is also relatively good. These lay the foundation for the digital detection and evaluation of the appearance quality of sawn timber. At the same time, by adjusting the relevant influence coefficient of the model, it can meet the needs of different tree species and different applications in order to achieve better evaluation result.

Key words: sawn timber, appearance quality, quantitative evaluation, digitization, evaluation model