| Summary: | 碩士 === 國立中興大學 === 生物產業機電工程學系 === 93 === Developing a set of on-line acquiring the external image for fruits was the purpose of this study. The wax-apple from Pingtung was applied in the experiment. Two color analog conveyer were used to acquire the top view and the side view. There are 8 parameters ( red area of top view, height, width, cone volume, color coordination, red, green, blue average gray of top view ) which were conducted to be analyzed from the acquired image, 6 parameters undependent were selected to be the input parameters of grading Back Propagation network model. Results of training and predicting test were shown as following: The classification rates could be reached 100% for training set and 98% for predicting set by applying the designed grading neural network model with two hidden layers and 6 neurons.
|
|---|
