# Filter Node

## Purpose & Use Cases

The `filter` node applies various image processing filters for enhancement, artistic effects, and image analysis. It provides optimized implementations of common filters including blur, sharpen, edge detection, emboss, and Gaussian operations.

**Real-World Applications:**
- **Photo Enhancement**: Sharpen portraits and improve image quality
- **Artistic Effects**: Create stylized images with emboss and edge effects
- **Quality Control**: Use edge detection to analyze product shapes and boundaries
- **Medical Imaging**: Enhance contrast and details in diagnostic images
- **Preprocessing**: Prepare images for AI analysis with noise reduction and enhancement

![Filter Demo](../../../assets/nodes/transform/filter-demo.gif)

## Input/Output Specification

### Inputs
- **Single Image**: Standard image object format
- **Image Array**: Array of image objects for batch filtering
- **Dynamic Parameters**: Filter strength and settings via message properties

### Outputs
- **Filtered Image**: Image with applied filter effects
- **Preserved Dimensions**: Output maintains original image dimensions
- **Format Options**: Raw image object or encoded file formats

## Configuration Options

### Input/Output Paths
- **Input From**: `msg.payload` (default), `flow.*`, `global.*`
- **Output To**: `msg.payload` (default), `flow.*`, `global.*`

### Filter Types

#### Blur Filter
- **Purpose**: Reduces image noise and creates soft focus effects
- **Strength**: Configurable blur radius
- **Use Cases**: Noise reduction, background softening, privacy protection

#### Sharpen Filter  
- **Purpose**: Enhances image details and edge definition
- **Strength**: Configurable sharpening intensity
- **Use Cases**: Photo enhancement, print preparation, detail emphasis

#### Edge Detection
- **Purpose**: Highlights edges and boundaries in images
- **Output**: Binary or grayscale edge maps
- **Use Cases**: Object detection, shape analysis, contour extraction

#### Emboss Filter
- **Purpose**: Creates 3D relief effect with depth appearance
- **Direction**: Configurable embossing direction
- **Use Cases**: Artistic effects, texture analysis, decorative processing

#### Gaussian Blur
- **Purpose**: Smooth blur with natural falloff
- **Sigma**: Configurable standard deviation for blur amount
- **Use Cases**: Background blur, noise reduction, preprocessing

### Filter Strength/Parameters
- **Range**: Varies by filter type
- **Sources**: Number, `msg.*`, `flow.*`, `global.*`
- **Dynamic**: Runtime adjustment via message properties

### Output Format Options
- **Raw**: Standard image object (fastest for processing chains)
- **JPEG**: Compressed with quality control
- **PNG**: Lossless preservation of filter effects
- **WebP**: Modern format with excellent compression

## Performance Notes

### C++ Backend Processing
- **Optimized Kernels**: Hand-tuned convolution implementations
- **OpenCV Integration**: Leverages OpenCV's optimized filter functions
- **Memory Efficient**: In-place processing where possible
- **Parallel Execution**: Multi-threaded processing for large images

### Filter-Specific Performance
- **Blur/Gaussian**: Linear time complexity with kernel size
- **Sharpen**: Fast single-pass convolution
- **Edge Detection**: Sobel or Canny edge detection algorithms
- **Emboss**: Single convolution pass with directional kernel