Color and texture monitoring based on computer vision during banana storage

The variation of color and texture during banana storage was studied based on computer vision. The images of green banana in first mature stage were acquired everyday during storage. After the binary operation, RGB and gray images were masked by the binary images to eliminate the background. Then, R, G and B average values were extracted from pre-processed color images as color indexes. The gray-level co-occurrence matrix generating rule of the irregular image was used to obtain these matrixes of pre-processed gray images. Three texture descriptors were extracted as texture indexes from these matrixes, namely energy, contrast and homogeneity, respectively. Results showed that the change curve of R average values in combination with that of G could describe banana surface condition before the sixth mature stage, and the curve of contrast and homogeneity could do that after the sixth mature stage.