# vq

**Repository Path**: williamzjc/vq

## Basic Information

- **Project Name**: vq
- **Description**: vector quantization
- **Primary Language**: Python
- **License**: MIT
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2022-04-20
- **Last Updated**: 2024-08-02

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# vq

#### Introduction

vector quantization for images

VQ as an encoder/transformer of images

### Installation

`git clone https://gitee.com/williamzjc/vq/`

### Usage

#### Import

`import vqtransform [as vq]` or `from vqtransform import vq`

#### Easy example

```python
im = Image.open('cloth.jpg')
t = VQTransformer(n_clusters=2)
t.fit(im)
im = t.reconstruct()
im.save('cloth.kmeans2.jpg')
```

#### Other examples

```python
    im = Image.open('lenna.jpg')

    block_size = 4, 4
    n_clusters = 3

    def demo(model, name='Encoder'):
        im0 = model.reconstruct()
        n_clusters = len(model.codebook_)

        import matplotlib.pyplot as plt
        fig, ax = plt.subplots(1,2)
        ax[0].imshow(im)
        ax[0].set_title('Original Image')
        ax[1].imshow(im0)
        ax[1].set_title(name)
        ax[0].axis('off')
        ax[1].axis('off')
        # fig.suptitle(f'{name} Demo (block-size: {block_size[0]} X {block_size[1]}, #clusters: {n_clusters})')
        plt.show()
    
    def demo_meanshift():
        ms = MeanShiftTransformer(block_size=block_size, max_iter=100)
        ms.fit(im)
        im0 = ms.reconstruct()
        ms.fit(im, with_pos=True)
        im1 = ms.reconstruct()

        import matplotlib.pyplot as plt
        fig, ax = plt.subplots(1,3)
        ax[0].imshow(im)
        ax[0].set_title('Original Image')
        ax[1].imshow(im0)
        ax[1].set_title('MeanShift')
        ax[2].imshow(im1)
        ax[2].set_title('MeanShift (with position)')
        for a in ax:
            a.axis('off')
        # fig.suptitle(f'Demo of MeanShift (block-size: {block_size[0]} X {block_size[1]})')
        plt.show()

    demo_meanshift()

    def _mask(im, block_size=(1,1), shape=None):
        """Transform an image to 2d array
        where block_size = (w, h)
        
        Arguments:
            im {Image} -- an image
        
        Keyword Arguments:
            block_size {tuple} -- a rectangle domain of an image (default: (1, 1))
            shape {tuple} -- size of small domain (default: {None})
                            calculated based on the size of the image by default
        
        Returns:
            tuple of a block matrix and its shape=(r, c)
        """
        w, h = block_size
        if shape is None:
            W, H = im.size
            c, cc = divmod(W, w)
            r, rr = divmod(H, h)
            shape = r, c
            im = im.resize((w*c, h*r))
        else:
            r, c = shape
            im = im.resize((0, 0, w*c, h*r))
        I = np.asarray(im, dtype=np.float64)
        shape = (r, c)
        # blocks <- concatenate the columns of the block matrix
        def _t(a):
            def _f(x):
                if np.all(x == (255,0,0)): return 0
                elif np.all(x == (0,255,0)): return 1
                elif np.all(x == (0,0,255)): return 2
                else: return -1
            return np.max(np.apply_along_axis(_f, 2, a))
        return np.array([[_t(I[i*h:(i+1)*h, j*w:(j+1)*w]) for j in range(c)] for i in range(r)])

    def semivq():
        svq = SemiVQTransformer(block_size=block_size, n_clusters=n_clusters)
        masked = Image.open('lenna-masked.jpg')
        mask = _mask(masked, block_size=block_size)
        svq.fit(im, mask=mask)
        im1 = svq.reconstruct()

        import matplotlib.pyplot as plt
        fig, ax = plt.subplots(2, 2)

        ax[0,0].imshow(im)
        ax[0,0].set_title('Original Image')
        ax[0,1].imshow(masked)
        ax[0,1].set_title('Masked Image')
        ax[1,0].imshow(im0)
        ax[1,0].set_title('VQ')
        ax[1,1].imshow(im1)
        ax[1,1].set_title('Semisupervised VQ')

        ax[0,0].axis('off')
        ax[0,1].axis('off')
        ax[1,0].axis('off')
        ax[1,1].axis('off')

        plt.show()
```

#### 参与贡献

1.  Fork 本仓库
2.  新建 Feat_xxx 分支
3.  提交代码
4.  新建 Pull Request


#### 特技

1.  使用 Readme\_XXX.md 来支持不同的语言，例如 Readme\_en.md, Readme\_zh.md
2.  Gitee 官方博客 [blog.gitee.com](https://blog.gitee.com)
3.  你可以 [https://gitee.com/explore](https://gitee.com/explore) 这个地址来了解 Gitee 上的优秀开源项目
4.  [GVP](https://gitee.com/gvp) 全称是 Gitee 最有价值开源项目，是综合评定出的优秀开源项目
5.  Gitee 官方提供的使用手册 [https://gitee.com/help](https://gitee.com/help)
6.  Gitee 封面人物是一档用来展示 Gitee 会员风采的栏目 [https://gitee.com/gitee-stars/](https://gitee.com/gitee-stars/)