Optimizing OCR Performance: An Investigation into Image Preprocessing Techniques

Optimizing OCR Performance: An Investigation into Image Preprocessing Techniques

Amaan Tamboli1, Prof. Archana Kadam2, Ayush Talegoankar3, Abhinav Tekam4, Bhushan Wakchaure5

1Amaan Tamboli Computer Engineering, PCCOE

2Archana Kadam Computer Engineering, PCCOE

3Ayush Talegaonkar Computer Engineering, PCCOE

4Abhinav Tekam Computer Engineering, PCCOE

5Bhushan Wakchaure Computer Engineering, PCCOE

---------------------------------------------------------------------***---------------------------------------------------------------------

Abstract - This study investigates the impact of various image processing techniques on the accuracy of text extraction by PaddleOCR. The research explores how inversion, noise reduction, erosion, dilation, and grayscaling can pre-process images to improve the performance of PaddleOCR. Initially, the OCR’s effectiveness is assessed using unprocessed images. Subsequently, each image processing method is applied individually to evaluate its contribution to the accuracy of text detection. The findings aim to identify the most beneficial preprocessing technique for enhancing the accuracy of PaddleOCR in text extraction tasks. This research has implications for the optimization of OCR technology in digitizing textual content from diverse image backgrounds.

Key Words:  OCR optimization, image preprocessing, paddleOCR, image to text