image-preprocessing

---

name: image-preprocessing description: Normalizes image histograms to preserve melanin features using Canvas API before analysis license: MIT compatibility: opencode metadata: audience: developers workflow: clinical-pipeline

What I do

I preprocess images to normalize histograms while preserving melanin features critical for skin analysis. I use the Canvas API for efficient pixel manipulation and contrast enhancement.

When to use me

Use this when:

• Calibration is complete and you need to prepare the image
• You need to enhance contrast for better lesion detection
• Histogram normalization is required before segmentation

Key Concepts

• Histogram Equalization: Spreads out pixel intensity values
• Melanin Preservation: Enhances features without losing skin pigmentation info
• Canvas API: Browser canvas for image manipulation
• image_preprocessed: State flag after preprocessing complete

Source Files

• services/vision.ts: Preprocessing implementation
• components/AnalysisIntake.tsx: Image upload component

Code Patterns

• Use Canvas API for pixel-level operations
• Apply histogram equalization that preserves melanin contrast
• Ensure output is suitable for segmentation algorithms

Operational Constraints

• Must preserve features critical for dermatological analysis
• Use tf.tidy() or explicit dispose() for any tensor operations
• Work in Canvas pixel space, not tensor space