The common edible shellfish recognition system using laser measurement

• Main Authors: CHEN,EI-LIN, 陳沂鈴
• Other Authors: WANG,MEI-JIN
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/b2za96

碩士 === 崑山科技大學 === 數位生活科技研究所 === 106 === Image recognition has been widely applied to sorts of fields. Nevertheless, the most common characteristics of image recognition, such as color, corner and texture, are liable to affected by illumination and the light of background leading to bias the recognized consequences.In recent years the technology of three-dimensional scanning has developed increasingly mature which could precisely measure the contour of fixed-shape objects.We take the three-dimensional contour as characteristics in the project. The comparative sample which will be used as for recognition was scanned every one mille meter from the leftmost to the rightmost. After position normalization and rotation, all of the data were saved in the three-dimensional object database.In order to shorten the testing time, we merely scanned two scanning lines while doing testing.There were twelve edible shellfishes used for experiment. Each category has individual thirty samples. We take leave-one-out for recognition mechanism that the rate of recognition could reach 66.66 percentage and the average of recognition time was 80 mille second using general notebook computer.Regardless of position and angle at which the shellfish was to be recognized, or whether the database incorporated pre-rotated three-dimensional shellfish sample data, the recognition rates were similar which indicating the future is robust. The three-dimensional scanning is a new option for characteristic capturing. Furthermore, we will do our best to continue increasing the rate and speed of recognition to improve the practicality in the future.