An answer card scoring method based on image recognition

Abstract

The invention discloses an answer card scoring method based on image recognition. The method comprises the following steps: firstly, the scanned answer card images are processed in gray scale and binarized to obtain binarized gray scale images; Then, the global positioning block is recognized, the front and back of the answer card are judged, and the image is rotated and rectified. Then, all the local positioning block pairs are identified, and screenshots are performed on each line of the questions. For the screenshots of the multiple choice questions, the filling and coating positions are identified, and the filling and coating answers are calculated according to the division proportion of the options. For the subjective screenshots, the scoring results are judged by the red HSV model. At last, compare the result data with the standard answer, and calculate the final score of the answer card. The invention reduces the hardware cost of the answer card identification, provides an identification method comprising a choice question and a subjective question, has high accuracy and fault tolerance rate, and improves the marking efficiency.

Description

technical field [0001] The invention belongs to the technical field of answer sheet recognition, in particular to an image recognition-based answer sheet scoring method. Background technique [0002] The answer sheet is a commonly used answering medium and means in the exam. Nowadays, the multiple choice part of the exam generally adopts the form of a standardized answer sheet, and the examinee uses a 2B pencil to fill in the answer sheet on the answer sheet. At present, the application of the answer sheet usually adopts a cursor reader device to scan the answer sheet and obtain the score data. Before the answer sheet is used, coordinate position anchor points must be set on the side or top of the answer sheet filling area, so that the cursor reader can be positioned at the position that needs to be recognized. The cursor reader has high requirements on the paper quality of the answer card, and ordinary printing paper cannot be recognized. The application cost of the answe...

AI Technical Summary

Problems solved by technology

The cursor reader has high requirements on the paper quality of the answer card, and ordinary printing paper cannot be recognized

The application cost of the answer sheet of the cursor reader is very high, and there are too many restrictions on its use. Therefore, the answer sheet of the cursor reader is mainly used in some large-scale examination occasions, and it is difficult for ordinary teachers to apply it to teaching activities such as daily exercises and tests. middle

[0003] At present, there are some technical means that can realize self-made answer sheets, which can be scanned and identified by ordinary scanners. However, due to objective factors such as uneven thickness of paper, different degrees of humidity of paper, static electricity, and inaccurate scanner equipment and paper feeding equipment, etc. Scanned answer sheet pictures often have problems such as position offset, tilt, and size distortion, which in turn lead to inaccurate answer sheet recognition and incorrect score data, which seriously affects the development of answer sheet technology

In addition, the current answer sheet recognition is limited to the recognition of objective questions, and there is a lack of recognition and statistics on the scores of subjective questions

Images