earticle

영어

This paper constructed a dataset of beef carcass images with a specific resolution, and evaluated its classification performance using artificial intelligence. The dataset was also analyzed based on color and texture features. Grade information by color texture was scored using linear or average values. To analyze the dataset with a higher level of precision, regression analysis was performed based on scores rather than grades. IOCLBP(Improved opponent color local binary pattern) was used for color texture analysis, and the correlation coefficient for each color was high in Blue-gray color space. The scores for each grade in Blue- gray space were first given linearly, and the average and exponential approximation of the actual scores were also evaluated. The evaluation involved scoring using the regression mean and the approximate exponential function from the linear score results. The correlation coefficients according to the scoring method were 0.641, 0.701, and 0.675 respectively. These results showed that scoring by average was most effective. Finally, the grades were divided into non-overlapping score ranges to check the color texture analysis scores of the training dataset. The regression coefficient of the dataset was 0.947, indicating its reliability.