# IntegralThreshold

**Repository Path**: woaichengyang/IntegralThreshold

## Basic Information

- **Project Name**: IntegralThreshold
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2020-06-10
- **Last Updated**: 2020-12-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# IntegralThreshold
Adaptive Thresholding Using the Integral Image.

Based on https://github.com/phryniszak/AdaptiveIntegralThresholding Converted to Python.</br>
For more details, please read this papaer [Adaptive Thresholding Using the Integral Image](https://www.researchgate.net/publication/220494200_Adaptive_Thresholding_using_the_Integral_Image).</br>

This method can preserve more image details as the table shows.  
<table>
<thead><tr><th>test image</th><th>OTSU Thresholding</th></tr></thead>
        <tr>
            <td><a href=""><img width="100%" style="max-width: 30%;max-height:30%;" alt="" src="https://github.com/Liang-yc/IntegralThreshold/blob/master/testimage.jpg" ></a></td>
            <td><a href=""><img width="100%" style="max-width: 30%;max-height:30%;" alt="" src="https://github.com/Liang-yc/IntegralThreshold/blob/master/otsu_results.jpg" ></a></td>
        </tr>
  <tr><th>Adaptive Thresholding</th><th>Adaptive Thresholding Using the Integral Image</th></tr></thead>
        <tr>
            <td><a href=""><img width="100%" style="max-width: 30%;max-height:30%;" alt="" src="https://github.com/Liang-yc/IntegralThreshold/blob/master/adaptive_results.jpg" ></a></td>
            <td><a href=""><img width="100%" style="max-width: 30%;max-height:30%;" alt="" src="https://github.com/Liang-yc/IntegralThreshold/blob/master/results.jpg" ></a></td>
        </tr>
</table>

The code is simple but the speed is insanely slow when the image is very large; and I use [numba](http://numba.pydata.org/) to speed up. The original code consumes more than 3 minutes for processing this test image while it only costs about 1 seconds with numba.