[中值过滤器，在Android上使用RenderScript

Question

我正在使用RenderScript进行图像处理的第一步。现在，我正在尝试使用它为Android实现中值过滤器。基本上，我尝试使用的是：它获取位图格式的图像，为每个像素及其8个邻居（没有边缘情况）运行一个3x3滤镜，计算每个颜色通道的中值并相应地设置输出像素。

我已经尝试过是否可以与像这样的简单反转渲染脚本一起使用：

uchar4 RS_KERNEL invert(uchar4 in, uint32_t x, uint32_t y) {
  uchar4 out = in;
  out.r = 255 - in.r;
  out.g = 255 - in.g;
  out.b = 255 - in.b;
  return out;
}

并且工作正常。但是，当我更改RenderScript以计算中间值时，我只得到了黑色图像输出。

这里是RenderScript代码：

#pragma version(1)
#pragma rs_fp_relaxed
#pragma rs java_package_name(com.google.android.GoogleCamera)

#include "rs_debug.rsh"

rs_allocation input;
uint32_t width;
uint32_t height;

static uchar medi(uchar colors[]) {
    //sorting just to half
    for(int i = 0; i < 5; i++){
        int minIndex = i;
        uchar minValue = colors[i];
        //loop until end
        for(int j = i+1; j < 9; j++){
            if (colors[j] < minValue) {
                minIndex = j;
                minValue = colors[j];
                //swap
                colors[j] = colors[i];
                colors[i] = minValue;
            }
        }
    }
    return colors[4];
}

uchar4 __attribute__((kernel)) median(uchar4 in, uint32_t x, uint32_t y) {
    uchar4 out = in;

    //edge
    if (x == 0 || x == width || y == 0 || y == height)
        return out;

    //all surronding pixels (TopLeft .. BottomRight)
    uchar4 tl = rsGetElementAt_uchar4(input, x-1, y+1);
    uchar4 tm = rsGetElementAt_uchar4(input, x, y+1);
    uchar4 tr = rsGetElementAt_uchar4(input, x+1, y+1);
    uchar4 l = rsGetElementAt_uchar4(input, x-1, y);
    uchar4 r = rsGetElementAt_uchar4(input, x+1, y);
    uchar4 bl = rsGetElementAt_uchar4(input, x-1, y-1);
    uchar4 bm = rsGetElementAt_uchar4(input, x, y-1);
    uchar4 br = rsGetElementAt_uchar4(input, x+1, y-1);

    //array for each color channel
    uchar reds[] = {tl.r, tm.r, tr.r, l.r, r.r, bl.r, bm.r, br.r, in.r};
    uchar greens[] = {tl.g, tm.g, tr.g, l.g, r.g, bl.g, bm.g, br.g, in.g};
    uchar blues[] = {tl.b, tm.b, tr.b, l.b, r.b, bl.b, bm.b, br.b, in.b};
    uchar alphas[] = {tl.a, tm.a, tr.a, l.a, r.a, bl.a, bm.a, br.a, in.a};

    //get median values
    out.r = medi(reds);
    out.g = medi(greens);
    out.b = medi(blues);
    out.a = medi(alphas);

    return out;
}

以防万一调用的Java部分：

private static Runnable runMedian(final Bitmap original, final File file) {
        return new Runnable() {
            @Override
            public void run() {
                //Create new bitmap
                Bitmap bitmap = original.copy(original.getConfig(), true);

                //Create renderscript
                RenderScript rs = RenderScript.create(CameraActivity.staticContext);

                //Create allocation from Bitmap
                Allocation allocationA = Allocation.createFromBitmap(rs, bitmap);

                //Create allocation with same type
                Allocation allocationB = Allocation.createTyped(rs, allocationA.getType());

                //Create script from rs file
                ScriptC_median medianScript = new ScriptC_median(rs);

                medianScript.set_input(allocationA);
                medianScript.set_width(original.getWidth());
                medianScript.set_height(original.getWidth());
                medianScript.forEach_median(allocationA, allocationB);

                //Copy script result into bitmap
                allocationB.copyTo(bitmap);

                //Destroy everything to free memory
                allocationA.destroy();
                allocationB.destroy();
                medianScript.destroy();
                rs.destroy();

                //write resulting bitmap to file
                writeFile(file, bitmap, "median");
            }
        };
    }

我希望知道更多的人可以帮助我。祝你好运>

我正在使用RenderScript进行图像处理的第一步。现在，我正在尝试使用它为Android实现中值过滤器。基本上，我尝试使用的方法是：它在位图中获取图像...

Answer 1

[您可能需要修改边缘的处理方式：索引x从0到width-1，因此边缘的宽度为1（且高度为1）。