OCR Math Solver
Computer vision pipeline using OCR to recognize handwritten mathematical expressions and solve them step-by-step. Combines image preprocessing, character segmentation, and symbolic computation.
Problem
Students solving math problems by hand have no quick way to verify their work. Typing expressions into a calculator is slow and error-prone. A system that reads handwritten math directly from a photo and produces step-by-step solutions bridges that gap.
Solution
OpenCV pipeline preprocesses input images: grayscale conversion, adaptive thresholding, morphological operations to isolate expression regions. Tesseract OCR extracts the mathematical text. SymPy parses and solves the expression symbolically, generating a step-by-step solution breakdown.
Architecture
OpenCV pipeline: grayscale conversion, Gaussian blur, adaptive thresholding, dilation/erosion to clean handwritten strokes and isolate expression bounding boxes.
Tesseract OCR (custom config for math characters) extracts the expression string from preprocessed image regions.
SymPy parses the OCR output into a symbolic expression tree and solves it — supporting arithmetic, algebra, and basic calculus.
Step-by-step solution displayed with intermediate simplification steps, final answer, and annotated image showing detected expression region.
Highlights
- OpenCV adaptive thresholding and morphological ops for handwriting cleanup.
- Tesseract OCR with custom config tuned for mathematical character sets.
- SymPy symbolic computation — supports algebra, arithmetic, and basic calculus.
- Step-by-step solution output with intermediate simplification steps.