史上最强的YUV转换RenderScript，支持转换成RGBA,BGRA，并且同时支持旋转，翻转，YUV数据格式支持NV21, YV12, YUV420_888, 分别对应Android Camera1API 以及Camera2 API 的YUV处理
? 这段时间在做一些Android Camera数据处理的相关工作，涉及到YUV转RGBA的事情，由于对性能要求极高我先后采用了多种方案，最开始的时候是试用OpenCV里面的方法，最后发现性能不是很理想，最后又使用了libyuv，这个库性能要比libyuv要好，对性能要求不是特别高的可以采用，这个库有一定的缺陷，YUV转换后图像还原度很不好，明显感觉到有色差，而且在Android手机上长时间运行，性能下降明显，最后没有办法了开始采用RenderScript，这里不得不吐槽一下Google 他仅仅提供了一个ScriptIntrinsicYuvToRGB，而且不支持YUV420_888,也不支持旋转，翻转，我呸渣男！ 还能怎么办自己动手吧！
话不多说直接上代码吧（里面涉及到很多YUV的知识我就不讲了）：
#pragma version(1)
#pragma rs java_package_name(com.luoye.bzyuv)
#pragma rs_fp_relaxedrs_allocation mInput;rs_allocation mInY;
rs_allocation mInU;
rs_allocation mInV;
rs_allocation mInNV21;
rs_allocation mInYV12;
int inWidth=0,inHeight=0;
int orientation=0;
bool flipY=false;
int uvPixelStride=1;static uchar4 yuvToRGBA4(uchar y, uchar u, uchar v) {short Y = ((short)y) - 16;short U = ((short)u) - 128;short V = ((short)v) - 128;short4 p;p.x = (Y * 298 + V * 409 + 128) >> 8;p.y = (Y * 298 - U * 100 - V * 208 + 128) >> 8;p.z = (Y * 298 + U * 516 + 128) >> 8;p.w = 255;if(p.x < 0) {p.x = 0;}if(p.x > 255) {p.x = 255;}if(p.y < 0) {p.y = 0;}if(p.y > 255) {p.y = 255;}if(p.z < 0) {p.z = 0;}if(p.z > 255) {p.z = 255;}return (uchar4){p.x, p.y, p.z, p.w};
}static float4 yuvToRGBA_f4(uchar y, uchar u, uchar v) {float4 yuv_U_values = {0.f, -0.392f * 0.003921569f, +2.02 * 0.003921569f, 0.f};float4 yuv_V_values = {1.603f * 0.003921569f, -0.815f * 0.003921569f, 0.f, 0.f};float4 color = (float)y * 0.003921569f;float4 fU = ((float)u) - 128.f;float4 fV = ((float)v) - 128.f;color += fU * yuv_U_values;color += fV * yuv_V_values;color = clamp(color, 0.f, 1.f);return color;
}void makeRef(rs_allocation ay, rs_allocation au, rs_allocation av, rs_allocation aout) {uint32_t w = rsAllocationGetDimX(ay);uint32_t h = rsAllocationGetDimY(ay);for (int y = 0; y < h; y++) {//rsDebug("y", y);for (int x = 0; x < w; x++) {int py = rsGetElementAt_uchar(ay, x, y);int pu = rsGetElementAt_uchar(au, x >> 1, y >> 1);int pv = rsGetElementAt_uchar(av, x >> 1, y >> 1);//rsDebug("py", py);//rsDebug(" u", pu);//rsDebug(" v", pv);uchar4 rgb = yuvToRGBA4(py, pu, pv);//rsDebug(" ", rgb);rsSetElementAt_uchar4(aout, rgb, x, y);}}
}void makeRef_f4(rs_allocation ay, rs_allocation au, rs_allocation av, rs_allocation aout) {uint32_t w = rsAllocationGetDimX(ay);uint32_t h = rsAllocationGetDimY(ay);for (int y = 0; y < h; y++) {//rsDebug("y", y);for (int x = 0; x < w; x++) {uchar py = rsGetElementAt_uchar(ay, x, y);uchar pu = rsGetElementAt_uchar(au, x >> 1, y >> 1);uchar pv = rsGetElementAt_uchar(av, x >> 1, y >> 1);//rsDebug("py", py);//rsDebug(" u", pu);//rsDebug(" v", pv);float4 rgb = yuvToRGBA_f4(py, pu, pv);//rsDebug(" ", rgb);rsSetElementAt_float4(aout, rgb, x, y);}}
}static void setLocation(int x,int y, int*targetX,int *targetY){int finalX=x;int finalY=y;if(orientation!=0){if(orientation==90){finalX=y;finalY=inHeight-1-x;if(flipY){finalX=inWidth-1-finalX;}}else if(orientation==180){finalX=inWidth - 1 - x;finalY=inHeight - 1 - y;if(flipY){finalY=inHeight-1-finalY;}}else if(orientation==270){finalX=inWidth - 1 - y;finalY=x;if(flipY){finalY=inHeight-1-finalY;}}}else{if(flipY){finalY=inHeight-1-finalY;}}*targetX=finalX;*targetY=finalY;
}uchar4 __attribute__((kernel)) yuv_420_888_2_rgba(uint32_t x, uint32_t y) {if(inWidth<=0||inHeight<=0){rsDebug("bz_inWidth<=0||inHeight<=0", inWidth,inHeight);return '0';}int targetX=x;int targetY=y;setLocation(x,y,&targetX,&targetY);uint uvIndex= uvPixelStride * (targetX/2) + inWidth*(targetY/2);uchar yps= rsGetElementAt_uchar(mInY,targetX,targetY);uchar u= rsGetElementAt_uchar(mInU,uvIndex);uchar v= rsGetElementAt_uchar(mInV,uvIndex);return yuvToRGBA4(yps, u, v);
}uchar4 __attribute__((kernel)) yuv_420_888_2_bgra(uint32_t x, uint32_t y) {if(inWidth<=0||inHeight<=0){rsDebug("bz_inWidth<=0||inHeight<=0", inWidth,inHeight);return '0';}int targetX=x;int targetY=y;setLocation(x,y,&targetX,&targetY);uint uvIndex= uvPixelStride * (targetX/2) + inWidth*(targetY/2);uchar yps= rsGetElementAt_uchar(mInY,targetX,targetY);uchar u= rsGetElementAt_uchar(mInU,uvIndex);uchar v= rsGetElementAt_uchar(mInV,uvIndex);uchar4 rgba= yuvToRGBA4(yps, u, v);return (uchar4){rgba.z, rgba.y, rgba.x,rgba.w};
}uchar4 __attribute__((kernel)) yuv_yv12_2_rgba(uint32_t x, uint32_t y) {if(inWidth<=0||inHeight<=0){rsDebug("bz_inWidth<=0||inHeight<=0", inWidth,inHeight);return '0';}int targetX=x;int targetY=y;setLocation(x,y,&targetX,&targetY);uint yIndex=targetX+inWidth*targetY;uint uIndex=inWidth*inHeight+ (targetX/2) + inWidth/2*(targetY/2);uint vIndex=inWidth*inHeight+inWidth*inHeight/4+ (targetX/2) + inWidth/2*(targetY/2);uchar yps= rsGetElementAt_uchar(mInYV12,yIndex);uchar u= rsGetElementAt_uchar(mInYV12,uIndex);uchar v= rsGetElementAt_uchar(mInYV12,vIndex);return yuvToRGBA4(yps, v, u);
}uchar4 __attribute__((kernel)) yuv_yv12_2_bgra(uint32_t x, uint32_t y) {if(inWidth<=0||inHeight<=0){rsDebug("bz_inWidth<=0||inHeight<=0", inWidth,inHeight);return '0';}int targetX=x;int targetY=y;setLocation(x,y,&targetX,&targetY);uint yIndex=targetX+inWidth*targetY;uint uIndex=inWidth*inHeight+ (targetX/2) + inWidth/2*(targetY/2);uint vIndex=inWidth*inHeight+inWidth*inHeight/4+ (targetX/2) + inWidth/2*(targetY/2);uchar yps= rsGetElementAt_uchar(mInYV12,yIndex);uchar u= rsGetElementAt_uchar(mInYV12,uIndex);uchar v= rsGetElementAt_uchar(mInYV12,vIndex);uchar4 rgba= yuvToRGBA4(yps, v, u);return (uchar4){rgba.z, rgba.y, rgba.x,rgba.w};
}uchar4 __attribute__((kernel)) yuv_nv21_2_rgba(uint32_t x, uint32_t y) {if(inWidth<=0||inHeight<=0){rsDebug("bz_inWidth<=0||inHeight<=0", inWidth,inHeight);return '0';}int targetX=x;int targetY=y;setLocation(x,y,&targetX,&targetY);uint yIndex=targetX+inWidth*targetY;uint uIndex=inWidth*inHeight+ 2*(targetX/2) + inWidth*(targetY/2);uint vIndex=inWidth*inHeight+ 2*(targetX/2) + inWidth*(targetY/2)+1;uchar yps= rsGetElementAt_uchar(mInNV21,yIndex);uchar u= rsGetElementAt_uchar(mInNV21,uIndex);uchar v= rsGetElementAt_uchar(mInNV21,vIndex);return yuvToRGBA4(yps, v, u);
}uchar4 __attribute__((kernel)) yuv_nv21_2_bgra(uint32_t x, uint32_t y) {if(inWidth<=0||inHeight<=0){rsDebug("bz_inWidth<=0||inHeight<=0", inWidth,inHeight);return '0';}int targetX=x;int targetY=y;setLocation(x,y,&targetX,&targetY);uint yIndex=targetX+inWidth*targetY;uint uIndex=inWidth*inHeight+ 2*(targetX/2) + inWidth*(targetY/2);uint vIndex=inWidth*inHeight+ 2*(targetX/2) + inWidth*(targetY/2)+1;uchar yps= rsGetElementAt_uchar(mInNV21,yIndex);uchar u= rsGetElementAt_uchar(mInNV21,uIndex);uchar v= rsGetElementAt_uchar(mInNV21,vIndex);uchar4 rgba= yuvToRGBA4(yps, v, u);return (uchar4){rgba.z, rgba.y, rgba.x,rgba.w};
}float4 __attribute__((kernel)) cvt_f4(uint32_t x, uint32_t y) {uchar py = rsGetElementAtYuv_uchar_Y(mInput, x, y);uchar pu = rsGetElementAtYuv_uchar_U(mInput, x, y);uchar pv = rsGetElementAtYuv_uchar_V(mInput, x, y);//rsDebug("py2", py);//rsDebug(" u2", pu);//rsDebug(" v2", pv);return rsYuvToRGBA_float4(py, pu, pv);
}
The supported methods are as follows：
- yuv_yv12_2_Bitmap
- yuv_yv12_2_RGBA
- yuv_yv12_2_BGRA
- yuv_nv21_2_Bitmap
- yuv_nv21_2_RGBA
- yuv_nv21_2_BGRA
- yuv420_2_Bitmap
- yuv420_2_RGBA
- yuv420_2_BGRA
原文与Demo地址：http://www.bzblog.online/wordpress/index.php/2020/01/19/yuvrenderscript/