BahasaMelayuMalaysiaEnglish (UK)

Ensemble Neural Network For Handwriting Recognition

Authors: Billy Looi Say Zun and Chin Kim on


In this research, an approach to recognize School of Engineering and Information Technology’s (SEIT) students’ ID and marks is presented. Initially, all SEIT lecturers have to key in the students’ ID and marks after marking all final examination papers. However, lecturers have to spend more than few hours to key in the marks as the number of marked papers involved 200 and above. Furthermore, the lecturers have to double check whether the marks keyed in the system is correct. Hence, a system is required in order to fasten the key in process and reduce the risk. The proposed system involved three major processes, (1) image acquisition, (2) image pre-processing, and (3) training and recognition. The raw images are captured with a Canon low quality scanner supported 300 dpi resolution. There are a total of 300 cover pages have been scanned and saved in .JPEG format. The scanned images are then apply for image pre-processing. In the pre-processing, the images are first cropped. Only the students ID and marks wrote by lecturer are retained. Then, the images are enhanced by removing the background colour.