Opengl ES之RGB转NV21

6 个月前 · 来自专栏 Opengl Es

思想觉悟

前言

在上一篇理论文章中我们介绍了YUV到RGB之间转换的几种公式与一些优化算法，今天我们再来介绍一下RGB到YUV的转换，顺便使用Opengl ES做个实践，将一张RGB的图片通过Shader 的方式转换YUV格式图，然后保存到本地。

可能有的童鞋会问，YUV转RGB是为了渲染显示，那么RGB转YUV的应用场景是什么？在做视频编码的时候我们可以使用MediaCodec搭配Surface就可以完成，貌似也没有用到RGB转YUV的功能啊，硬编码没有用到，那么软编码呢？一般我们做视频编码的时候都是硬编码优先，软编码兜底的原则，在遇到一些硬编码不可用的情况下可能就需要用到x264库进行软编码了，而此时RGB转YUV可能就派上用场啦。

RGB到YUV的转换公式

在前面 Opengl ES之YUV数据渲染 一文中我们介绍过YUV的几种兼容标准，下面我们看看RGB到YUV的转换公式：

RGB 转 BT.601 YUV

Y  =  0.257R + 0.504G + 0.098B + 16
Cb = -0.148R - 0.291G + 0.439B + 128
Cr =  0.439R - 0.368G - 0.071B + 128

RGB 转 BT.709 YUV

Y  =  0.183R + 0.614G + 0.062B + 16
Cb = -0.101R - 0.339G + 0.439B + 128
Cr =  0.439R - 0.399G - 0.040B + 128

或者也可以使用矩阵运算的方式进行转换，更加的便捷：

RGB转YUV

先说一下RGB转YUV的过程，先将RGB数据按照公式转换为YUV数据，然后将YUV数据按照RGBA进行排布，这一步的目的是为了后续数据读取，最后使用 glReadPixels 读取YUV数据。

而对于OpenGL ES来说，目前它输入只认RGBA、lumiance、luminace alpha这几个格式，输出大多数实现只认RGBA格式，因此输出的数据格式虽然是YUV格式，但是在存储时我们仍然要按照RGBA方式去访问texture数据。

以NV21的YUV数据为例，它的内存大小为 width x height * 3 / 2 。如果是RGBA的格式存储的话，占用的内存空间大小是 width x height x 4 （因为 RGBA 一共4个通道）。很显然它们的内存大小是对不上的，那么该如何调整Opengl buffer的大小让RGBA的输出能对应上YUV的输出呢？我们可以设计输出的宽为 width / 4 ，高为 height * 3 / 2 即可。

为什么是这样的呢？虽然我们的目的是将RGB转换成YUV，但是我们的输入和输出时读取的类型GLenum是依然是RGBA，也就是说：width x height x 4 = (width / 4) x (height * 3 / 2) * 4

而YUV数据在内存中的分布以下这样子的：

width / 4
|--------------|
|              |
|              | h
|      Y       |
|--------------|            
|   U   |  V   |
|       |      |  h / 2
|--------------|

那么上面的排序如果进行了归一化之后呢，就变成了下面这样子了：

(0,0) width / 4  (1,0)
|--------------|
|              |
|              |  h
|      Y       |
|--------------|  (1,2/3)          
|   U   |  V   |
|       |      |  h / 2
|--------------|
(0,1)           (1,1)

从上面的排布可以看出看出，在纹理坐标 y < (2/3) 时，需要完成一次对整个纹理的采样，用于生成Y数据，当纹理坐标 y > (2/3) 时，同样需要再进行一次对整个纹理的采样，用于生成UV的数据。同时还需要将我们的视窗设置为 glViewport(0, 0, width / 4, height * 1.5);

由于视口宽度设置为原来的 1/4 ，可以简单的认为相对于原来的图像每隔4个像素做一次采样，由于我们生成Y数据是要对每一个像素都进行采样，所以还需要进行3次偏移采样。

同理，生成对于UV数据也需要进行3次额外的偏移采样。

在着色器中offset变量需要设置为一个归一化之后的值： 1.0/width , 按照原理图，在纹理坐标 y < (2/3) 范围，一次采样（加三次偏移采样）4 个 RGBA 像素（R,G,B,A）生成 1 个（Y0,Y1,Y2,Y3），整个范围采样结束时填充好 width*height 大小的缓冲区；当纹理坐标 y > (2/3) 范围，一次采样（加三次偏移采样）4 个 RGBA 像素（R,G,B,A）生成 1 个（V0,U0,V0,U1），又因为 UV 缓冲区的高度为 height/2 ，VU plane 在垂直方向的采样是隔行进行，整个范围采样结束时填充好 width*height/2 大小的缓冲区。

主要代码

RGBtoYUVOpengl.cpp
#include "../utils/Log.h"
#include "RGBtoYUVOpengl.h"
// 顶点着色器
static const char *ver = "#version 300 es\n"
                         "in vec4 aPosition;\n"
                         "in vec2 aTexCoord;\n"
                         "out vec2 v_texCoord;\n"
                         "void main() {\n"
                         "  v_texCoord = aTexCoord;\n"
                         "  gl_Position = aPosition;\n"
// 片元着色器
static const char *fragment = "#version 300 es\n"
                              "precision mediump float;\n"
                              "in vec2 v_texCoord;\n"
                              "layout(location = 0) out vec4 outColor;\n"
                              "uniform sampler2D s_TextureMap;\n"
                              "uniform float u_Offset;\n"
                              "const vec3 COEF_Y = vec3(0.299, 0.587, 0.114);\n"
                              "const vec3 COEF_U = vec3(-0.147, -0.289, 0.436);\n"
                              "const vec3 COEF_V = vec3(0.615, -0.515, -0.100);\n"
                              "const float UV_DIVIDE_LINE = 2.0 / 3.0;\n"
                              "void main(){\n"
                              "    vec2 texelOffset = vec2(u_Offset, 0.0);\n"
                              "    if (v_texCoord.   y <= UV_DIVIDE_LINE) {\n"
                              "        vec2 texCoord = vec2(v_texCoord.   x, v_texCoord.   y * 3.0 / 2.0);\n"
                              "        vec4 color0 = texture(s_TextureMap, texCoord);\n"
                              "        vec4 color1 = texture(s_TextureMap, texCoord + texelOffset);\n"
                              "        vec4 color2 = texture(s_TextureMap, texCoord + texelOffset * 2.0);\n"
                              "        vec4 color3 = texture(s_TextureMap, texCoord + texelOffset * 3.0);\n"
                              "        float y0 = dot(color0.   rgb, COEF_Y);\n"
                              "        float y1 = dot(color1.   rgb, COEF_Y);\n"
                              "        float y2 = dot(color2.   rgb, COEF_Y);\n"
                              "        float y3 = dot(color3.   rgb, COEF_Y);\n"
                              "        outColor = vec4(y0, y1, y2, y3);\n"
                              "    } else {\n"
                              "        vec2 texCoord = vec2(v_texCoord.x, (v_texCoord.y - UV_DIVIDE_LINE) * 3.0);\n"
                              "        vec4 color0 = texture(s_TextureMap, texCoord);\n"
                              "        vec4 color1 = texture(s_TextureMap, texCoord + texelOffset);\n"
                              "        vec4 color2 = texture(s_TextureMap, texCoord + texelOffset * 2.0);\n"
                              "        vec4 color3 = texture(s_TextureMap, texCoord + texelOffset * 3.0);\n"
                              "        float v0 = dot(color0.   rgb, COEF_V) + 0.5;\n"
                              "        float u0 = dot(color1.   rgb, COEF_U) + 0.5;\n"
                              "        float v1 = dot(color2.   rgb, COEF_V) + 0.5;\n"
                              "        float u1 = dot(color3.   rgb, COEF_U) + 0.5;\n"
                              "        outColor = vec4(v0, u0, v1, u1);\n"
                              "    }\n"
// 使用绘制两个三角形组成一个矩形的形式(三角形带)
// 第一第二第三个点组成一个三角形，第二第三第四个点组成一个三角形
const static GLfloat VERTICES[] = {
        1.0f,-1.0f, // 右下
        1.0f,1.0f, // 右上
        -1.0f,-1.0f, // 左下
        -1.0f,1.0f // 左上
// FBO贴图纹理坐标（参考手机屏幕坐标系统，原点在左下角）
// 注意坐标不要错乱
const static GLfloat TEXTURE_COORD[] = {
        1.0f,0.0f, // 右下
        1.0f,1.0f, // 右上
        0.0f,0.0f, // 左下
        0.0f,1.0f // 左上
RGBtoYUVOpengl::RGBtoYUVOpengl() {
    initGlProgram(ver,fragment);
    positionHandle = glGetAttribLocation(program,"aPosition");
    textureHandle = glGetAttribLocation(program,"aTexCoord");
    textureSampler = glGetUniformLocation(program,"s_TextureMap");
    u_Offset = glGetUniformLocation(program,"u_Offset");
    LOGD("program:%d",program);
    LOGD("positionHandle:%d",positionHandle);
    LOGD("textureHandle:%d",textureHandle);
    LOGD("textureSample:%d",textureSampler);
    LOGD("u_Offset:%d",u_Offset);
RGBtoYUVOpengl::~RGBtoYUVOpengl() noexcept {
void RGBtoYUVOpengl::fboPrepare() {
    glGenTextures(1, &fboTextureId);
    // 绑定纹理
    glBindTexture(GL_TEXTURE_2D, fboTextureId);
    // 为当前绑定的纹理对象设置环绕、过滤方式
    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
    glBindTexture(GL_TEXTURE_2D, GL_NONE);
    glGenFramebuffers(1,&fboId);
    glBindFramebuffer(GL_FRAMEBUFFER,fboId);
    // 绑定纹理
    glBindTexture(GL_TEXTURE_2D,fboTextureId);
    glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, fboTextureId, 0);
    // 这个纹理是多大的？
    glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, imageWidth / 4, imageHeight * 1.5, 0, GL_RGBA, GL_UNSIGNED_BYTE, nullptr);
    // 检查FBO状态
    if (glCheckFramebufferStatus(GL_FRAMEBUFFER)!= GL_FRAMEBUFFER_COMPLETE) {
        LOGE("FBOSample::CreateFrameBufferObj glCheckFramebufferStatus status != GL_FRAMEBUFFER_COMPLETE");
    // 解绑
    glBindTexture(GL_TEXTURE_2D, GL_NONE);
    glBindFramebuffer(GL_FRAMEBUFFER, GL_NONE);
// 渲染逻辑
void RGBtoYUVOpengl::onDraw() {
    // 绘制到FBO上去
    // 绑定fbo
    glBindFramebuffer(GL_FRAMEBUFFER, fboId);
    glPixelStorei(GL_UNPACK_ALIGNMENT,1);
    // 设置视口大小
    glViewport(0, 0,imageWidth / 4, imageHeight * 1.5);
    glClearColor(0.0f, 1.0f, 0.0f, 1.0f);
    glClear(GL_COLOR_BUFFER_BIT);
    glUseProgram(program);
    // 激活纹理
    glActiveTexture(GL_TEXTURE2);
    glUniform1i(textureSampler, 2);
    // 绑定纹理
    glBindTexture(GL_TEXTURE_2D, textureId);
    // 设置偏移
    float texelOffset = (float) (1.f / (float) imageWidth);
    glUniform1f(u_Offset,texelOffset);
     * size 几个数字表示一个点，显示是两个数字表示一个点
     * normalized 是否需要归一化，不用，这里已经归一化了
     * stride 步长，连续顶点之间的间隔，如果顶点直接是连续的，也可填0
    // 启用顶点数据
    glEnableVertexAttribArray(positionHandle);
    glVertexAttribPointer(positionHandle,2,GL_FLOAT,GL_FALSE,0,VERTICES);
    // 纹理坐标
    glEnableVertexAttribArray(textureHandle);
    glVertexAttribPointer(textureHandle,2,GL_FLOAT,GL_FALSE,0,TEXTURE_COORD);
    // 4个顶点绘制两个三角形组成矩形
    glDrawArrays(GL_TRIANGLE_STRIP,0,4);
    glUseProgram(0);
    // 禁用顶点
    glDisableVertexAttribArray(positionHandle);
    if(nullptr != eglHelper){
        eglHelper->swapBuffers();
    glBindTexture(GL_TEXTURE_2D, 0);
    // 解绑fbo
    glBindFramebuffer(GL_FRAMEBUFFER, 0);
// 设置RGB图像数据
void RGBtoYUVOpengl::setPixel(void *data, int width, int height, int length) {
    LOGD("texture setPixel");
    imageWidth = width;
    imageHeight = height;
    // 准备fbo
    fboPrepare();
    glGenTextures(1, &textureId);
    // 激活纹理，注意以下这个两句是搭配的，glActiveTexture激活的是那个纹理，就设置的sampler2D是那个
    // 默认是0，如果不是0的话，需要在onDraw的时候重新激活一下？
//    glActiveTexture(GL_TEXTURE0);
//    glUniform1i(textureSampler, 0);
// 例如，一样的
    glActiveTexture(GL_TEXTURE2);
    glUniform1i(textureSampler, 2);
    // 绑定纹理
    glBindTexture(GL_TEXTURE_2D, textureId);
    // 为当前绑定的纹理对象设置环绕、过滤方式
    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT);
    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT);
    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
    glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, width, height, 0, GL_RGBA, GL_UNSIGNED_BYTE, data);
    // 生成mip贴图
    glGenerateMipmap(GL_TEXTURE_2D);
    // 解绑定
    glBindTexture(GL_TEXTURE_2D, 0);
// 读取渲染后的YUV数据
void RGBtoYUVOpengl::readYUV(uint8_t **data, int *width, int *height) {
    // 从fbo中读取
    // 绑定fbo
    *width = imageWidth;
    *height = imageHeight;
    glBindFramebuffer(GL_FRAMEBUFFER, fboId);
    glBindTexture(GL_TEXTURE_2D, fboTextureId);
    glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0,
            GL_TEXTURE_2D, fboTextureId, 0);
    *data = new uint8_t[imageWidth * imageHeight * 3 / 2];
    glReadPixels(0, 0, imageWidth / 4, imageHeight * 1.5, GL_RGBA, GL_UNSIGNED_BYTE, *data);
    glBindTexture(GL_TEXTURE_2D, 0);
    // 解绑fbo
    glBindFramebuffer(GL_FRAMEBUFFER, 0);

下面是Activity的主要代码逻辑：

public class RGBToYUVActivity extends AppCompatActivity {
    protected MyGLSurfaceView myGLSurfaceView;
    @Override
    protected void onCreate(@Nullable Bundle savedInstanceState) {
        super.onCreate(savedInstanceState);
        setContentView(R.layout.activity_rgb_to_yuv);
        myGLSurfaceView = findViewById(R.id.my_gl_surface_view);
        myGLSurfaceView.setOpenGlListener(new MyGLSurfaceView.OnOpenGlListener() {
            @Override
            public BaseOpengl onOpenglCreate() {
                return new RGBtoYUVOpengl();
            @Override
            public Bitmap requestBitmap() {
                BitmapFactory.Options options = new BitmapFactory.Options();
                options.inScaled = false;
                return BitmapFactory.decodeResource(getResources(),R.mipmap.ic_smile,options);
            @Override
            public void readPixelResult(byte[] bytes) {
                if (null != bytes) {
            // 也就是RGBtoYUVOpengl::readYUV读取到结果数据回调
            @Override
            public void readYUVResult(byte[] bytes) {
                if (null != bytes) {
                    String fileName = System.currentTimeMillis() + ".yuv";
                    File fileParent = getFilesDir();
                    if (!fileParent.exists()) {
                        fileParent.mkdirs();
                    FileOutputStream fos = null;
                    try {
                        File file = new File(fileParent, fileName);
                        fos = new FileOutputStream(file);
                        fos.write(bytes,0,bytes.length);
                        fos.flush();
                        fos.close();
                        Toast.makeText(RGBToYUVActivity.this, "YUV图片保存成功" + file.getAbsolutePath(), Toast.LENGTH_LONG).show();
                    } catch (Exception e) {
                        Log.v("fly_learn_opengl", "图片保存异常：" + e.getMessage());
                        Toast.makeText(RGBToYUVActivity.this, "YUV图片保存失败", Toast.LENGTH_LONG).show();
        Button button = findViewById(R.id.bt_rgb_to_yuv);
        button.setOnClickListener(new View.OnClickListener() {
            @Override
            public void onClick(View view) {
                myGLSurfaceView.readYuvData();
        ImageView iv_rgb = findViewById(R.id.iv_rgb);
        iv_rgb.setImageResource(R.mipmap.ic_smile);

以下是自定义SurfaceView的代码：

public class MyGLSurfaceView extends SurfaceView implements SurfaceHolder.Callback {
    private final static int MSG_CREATE_GL = 101;
    private final static int MSG_CHANGE_GL = 102;
    private final static int MSG_DRAW_GL = 103;
    private final static int MSG_DESTROY_GL = 104;
    private final static int MSG_READ_PIXEL_GL = 105;
    private final static int MSG_UPDATE_BITMAP_GL = 106;
    private final static int MSG_UPDATE_YUV_GL = 107;
    private final static int MSG_READ_YUV_GL = 108;
    public BaseOpengl baseOpengl;
    private OnOpenGlListener onOpenGlListener;
    private HandlerThread handlerThread;
    private Handler renderHandler;
    public int surfaceWidth;
    public int surfaceHeight;
    public MyGLSurfaceView(Context context) {
        this(context,null);
    public MyGLSurfaceView(Context context, AttributeSet attrs) {
        super(context, attrs);
        getHolder().addCallback(this);
        handlerThread = new HandlerThread("RenderHandlerThread");
        handlerThread.start();
        renderHandler = new Handler(handlerThread.getLooper()){
            @Override
            public void handleMessage(@NonNull Message msg) {
                switch (msg.what){
                    case MSG_CREATE_GL:
                        baseOpengl = onOpenGlListener.onOpenglCreate();
                        Surface surface = (Surface) msg.obj;
                        if(null != baseOpengl){
                            baseOpengl.surfaceCreated(surface);
                            Bitmap bitmap = onOpenGlListener.requestBitmap();
                            if(null != bitmap){
                                baseOpengl.setBitmap(bitmap);
                        break;
                    case MSG_CHANGE_GL:
                        if(null != baseOpengl){
                            Size size = (Size) msg.obj;
                            baseOpengl.surfaceChanged(size.getWidth(),size.getHeight());
                        break;
                    case MSG_DRAW_GL:
                        if(null != baseOpengl){
                            baseOpengl.onGlDraw();
                        break;
                    case MSG_READ_PIXEL_GL:
                        if(null != baseOpengl){
                           byte[] bytes = baseOpengl.readPixel();
                           if(null != bytes && null != onOpenGlListener){
                               onOpenGlListener.readPixelResult(bytes);
                        break;
                    case MSG_READ_YUV_GL:
                        if(null != baseOpengl){
                            byte[] bytes = baseOpengl.readYUVResult();
                            if(null != bytes && null != onOpenGlListener){
                                onOpenGlListener.readYUVResult(bytes);
                        break;
                    case MSG_UPDATE_BITMAP_GL:
                        if(null != baseOpengl){
                            Bitmap bitmap = onOpenGlListener.requestBitmap();
                            if(null != bitmap){
                                baseOpengl.setBitmap(bitmap);
                                baseOpengl.onGlDraw();
                        break;
                    case MSG_UPDATE_YUV_GL:
                        if(null != baseOpengl){
                            YUVBean yuvBean = (YUVBean) msg.obj;
                            if(null != yuvBean){
                                baseOpengl.setYuvData(yuvBean.getyData(),yuvBean.getUvData(),yuvBean.getWidth(),yuvBean.getHeight());
                                baseOpengl.onGlDraw();
                        break;
                    case MSG_DESTROY_GL:
                        if(null != baseOpengl){
                            baseOpengl.surfaceDestroyed();
                        break;
    public void setOpenGlListener(OnOpenGlListener listener) {
        this.onOpenGlListener = listener;
    @Override
    public void surfaceCreated(@NonNull SurfaceHolder surfaceHolder) {
        Message message = Message.obtain();
        message.what = MSG_CREATE_GL;
        message.obj = surfaceHolder.getSurface();
        renderHandler.sendMessage(message);
    @Override
    public void surfaceChanged(@NonNull SurfaceHolder surfaceHolder, int i, int w, int h) {
        Message message = Message.obtain();
        message.what = MSG_CHANGE_GL;
        message.obj = new Size(w,h);
        renderHandler.sendMessage(message);
        Message message1 = Message.obtain();
        message1.what = MSG_DRAW_GL;
        renderHandler.sendMessage(message1);
        surfaceWidth = w;
        surfaceHeight = h;
    @Override
    public void surfaceDestroyed(@NonNull SurfaceHolder surfaceHolder) {
        Message message = Message.obtain();
        message.what = MSG_DESTROY_GL;
        renderHandler.sendMessage(message);
    public void readGlPixel(){
        Message message = Message.obtain();
        message.what = MSG_READ_PIXEL_GL;
        renderHandler.sendMessage(message);
    public void readYuvData(){
        Message message = Message.obtain();
        message.what = MSG_READ_YUV_GL;
        renderHandler.sendMessage(message);
    public void updateBitmap(){
        Message message = Message.obtain();
        message.what = MSG_UPDATE_BITMAP_GL;
        renderHandler.sendMessage(message);
    public void setYuvData(byte[] yData,byte[] uvData,int width,int height){
        Message message = Message.obtain();
        message.what = MSG_UPDATE_YUV_GL;
        message.obj = new YUVBean(yData,uvData,width,height);
        renderHandler.sendMessage(message);
    public void release(){
        // todo 主要线程同步问题，当心surfaceDestroyed还没有执行到，但是就被release了，那就内存泄漏了
        if(null != baseOpengl){
            baseOpengl.release();
    public void requestRender(){
        Message message = Message.obtain();
        message.what = MSG_DRAW_GL;
        renderHandler.sendMessage(message);
    public interface OnOpenGlListener{
        BaseOpengl onOpenglCreate();
        Bitmap requestBitmap();
        void readPixelResult(byte[] bytes);
        void readYUVResult(byte[] bytes);

BaseOpengl的java代码：

public class BaseOpengl {
    public static final int YUV_DATA_TYPE_NV12 = 0;
    public static final int YUV_DATA_TYPE_NV21 = 1;
    // 三角形
    public static final int DRAW_TYPE_TRIANGLE = 0;
    // 四边形
    public static final int DRAW_TYPE_RECT = 1;
    // 纹理贴图
    public static final int DRAW_TYPE_TEXTURE_MAP = 2;
    // 矩阵变换
    public static final int DRAW_TYPE_MATRIX_TRANSFORM = 3;
    // VBO/VAO
    public static final int DRAW_TYPE_VBO_VAO = 4;
    // EBO
    public static final int DRAW_TYPE_EBO_IBO = 5;
    // FBO
    public static final int DRAW_TYPE_FBO = 6;
    // PBO
    public static final int DRAW_TYPE_PBO = 7;
    // YUV  nv12与nv21渲染
    public static final int DRAW_YUV_RENDER = 8;
    // 将rgb图像转换城nv21
    public static final int DRAW_RGB_TO_YUV = 9;
    public long glNativePtr;
    protected EGLHelper eglHelper;
    protected int drawType;
    public BaseOpengl(int drawType) {
        this.drawType = drawType;
        this.eglHelper = new EGLHelper();
    public void surfaceCreated(Surface surface) {
        Log.v("fly_learn_opengl","------------surfaceCreated:" + surface);
        eglHelper.surfaceCreated(surface);
    public void surfaceChanged(int width, int height) {
        Log.v("fly_learn_opengl","------------surfaceChanged:" + Thread.currentThread());
        eglHelper.surfaceChanged(width,height);
    public void surfaceDestroyed() {
        Log.v("fly_learn_opengl","------------surfaceDestroyed:" + Thread.currentThread());
        eglHelper.surfaceDestroyed();
    public void release(){
        if(glNativePtr != 0){
            n_free(glNativePtr,drawType);
            glNativePtr = 0;
    public void onGlDraw(){
        Log.v("fly_learn_opengl","------------onDraw:" + Thread.currentThread());
        if(glNativePtr == 0){
            glNativePtr = n_gl_nativeInit(eglHelper.nativePtr,drawType);
        if(glNativePtr != 0){
            n_onGlDraw(glNativePtr,drawType);
    public void setBitmap(Bitmap bitmap){
        if(glNativePtr == 0){
            glNativePtr = n_gl_nativeInit(eglHelper.nativePtr,drawType);
        if(glNativePtr != 0){
            n_setBitmap(glNativePtr,bitmap);
    public void setYuvData(byte[] yData,byte[] uvData,int width,int height){
        if(glNativePtr != 0){
            n_setYuvData(glNativePtr,yData,uvData,width,height,drawType);
    public void setMvpMatrix(float[] mvp){
        if(glNativePtr == 0){
            glNativePtr = n_gl_nativeInit(eglHelper.nativePtr,drawType);
        if(glNativePtr != 0){
            n_setMvpMatrix(glNativePtr,mvp);
    public byte[] readPixel(){
        if(glNativePtr != 0){
            return n_readPixel(glNativePtr,drawType);
        return null;
    public byte[] readYUVResult(){
        if(glNativePtr != 0){
            return n_readYUV(glNativePtr,drawType);
        return null;
    // 绘制
    private native void n_onGlDraw(long ptr,int drawType);
    private native void n_setMvpMatrix(long ptr,float[] mvp);
    private native void n_setBitmap(long ptr,Bitmap bitmap);
    protected native long n_gl_nativeInit(long eglPtr,int drawType);
    private native void n_free(long ptr,int drawType);