目录

前言

插值算法

双线性插值算法

代码示例

前言

相机生成的图像一般是bayer阵列的raw图，这时候如果转换成我们需要的rgb图像，就需要对应的bayer插值算法将丢失的rgb信息通过算法补齐

插值算法

常见的插值算法一般有如下几种

• 邻近插值算法
• 双线性插值算法
• 高质量线性插值算法
• 边缘感应算法
• VNG算法(这里暂不介绍）
• AHD算法(这里暂不介绍）

双线性插值算法

双线性插值是再bayer转rgb中最常用的方法。由于直接取旁边像素容易使图像马赛克话，因此，双线性插值通过取其相邻像素的平均值来内插每个通道中的遗漏值，比如蓝色通道，缺少了红色和绿色，就以周边的红色和绿色来算平均值，

选取当前需要插值的像素点，以3*3矩阵去分割，缺少的像素就按下图算平均值即可

代码示例

void ClearBorders(uint8_t *rgb, int sx, int sy, int w)
{int i, j;// black edges are added with a width w:i = 3 * sx * w - 1;j = 3 * sx * sy - 1;while (i >= 0) {rgb[i--] = 0;rgb[j--] = 0;}int low = sx * (w - 1) * 3 - 1 + w * 3;i = low + sx * (sy - w * 2 + 1) * 3;while (i > low) {j = 6 * w;while (j > 0) {rgb[i--] = 0;j--;}i -= (sx - 2 * w) * 3;}
}int bayer_Bilinear(const uint8_t *B2R_RESTRICT bayer, uint8_t *B2R_RESTRICT rgb, int sx, int sy, int tile)
{const int bayerStep = sx;const int rgbStep = 3 * sx;int width = sx;int height = sy;/*the two letters  of the OpenCV name are respectivelythe 4th and 3rd letters from the blinky name,and we also have to switch R and B (OpenCV is BGR)CV_BayerBG2BGR <-> DC1394_COLOR_FILTER_BGGRCV_BayerGB2BGR <-> DC1394_COLOR_FILTER_GBRGCV_BayerGR2BGR <-> DC1394_COLOR_FILTER_GRBGint blue = tile == CV_BayerBG2BGR || tile == CV_BayerGB2BGR ? -1 : 1;int start_with_green = tile == CV_BayerGB2BGR || tile == CV_BayerGR2BGR;*/int blue = tile == DC1394_COLOR_FILTER_BGGR|| tile == DC1394_COLOR_FILTER_GBRG ? -1 : 1;int start_with_green = tile == DC1394_COLOR_FILTER_GBRG|| tile == DC1394_COLOR_FILTER_GRBG;if ((tile>DC1394_COLOR_FILTER_MAX)||(tile> 1; */t0 = (bayer[1] + bayer[bayerStep * 2 + 1] + 1) >> 1;t1 = (bayer[bayerStep] + bayer[bayerStep + 2] + 1) >> 1;rgb[-blue] = (uint8_t) t0;rgb[0] = bayer[bayerStep + 1];rgb[blue] = (uint8_t) t1;bayer++;rgb += 3;}if (blue > 0) {for (; bayer <= bayerEnd - 2; bayer += 2, rgb += 6) {t0 = (bayer[0] + bayer[2] + bayer[bayerStep * 2] +bayer[bayerStep * 2 + 2] + 2) >> 2;t1 = (bayer[1] + bayer[bayerStep] +bayer[bayerStep + 2] + bayer[bayerStep * 2 + 1] +2) >> 2;rgb[-1] = (uint8_t) t0;rgb[0] = (uint8_t) t1;rgb[1] = bayer[bayerStep + 1];t0 = (bayer[2] + bayer[bayerStep * 2 + 2] + 1) >> 1;t1 = (bayer[bayerStep + 1] + bayer[bayerStep + 3] +1) >> 1;rgb[2] = (uint8_t) t0;rgb[3] = bayer[bayerStep + 2];rgb[4] = (uint8_t) t1;}} else {for (; bayer <= bayerEnd - 2; bayer += 2, rgb += 6) {t0 = (bayer[0] + bayer[2] + bayer[bayerStep * 2] +bayer[bayerStep * 2 + 2] + 2) >> 2;t1 = (bayer[1] + bayer[bayerStep] +bayer[bayerStep + 2] + bayer[bayerStep * 2 + 1] +2) >> 2;rgb[1] = (uint8_t) t0;rgb[0] = (uint8_t) t1;rgb[-1] = bayer[bayerStep + 1];t0 = (bayer[2] + bayer[bayerStep * 2 + 2] + 1) >> 1;t1 = (bayer[bayerStep + 1] + bayer[bayerStep + 3] +1) >> 1;rgb[4] = (uint8_t) t0;rgb[3] = bayer[bayerStep + 2];rgb[2] = (uint8_t) t1;}}if (bayer < bayerEnd) {t0 = (bayer[0] + bayer[2] + bayer[bayerStep * 2] +bayer[bayerStep * 2 + 2] + 2) >> 2;t1 = (bayer[1] + bayer[bayerStep] +bayer[bayerStep + 2] + bayer[bayerStep * 2 + 1] +2) >> 2;rgb[-blue] = (uint8_t) t0;rgb[0] = (uint8_t) t1;rgb[blue] = bayer[bayerStep + 1];bayer++;rgb += 3;}bayer -= width;rgb -= width * 3;blue = -blue;start_with_green = !start_with_green;}return 0;
}

本文来自互联网用户投稿，文章观点仅代表作者本人，不代表本站立场，不承担相关法律责任。如若转载，请注明出处。 如若内容造成侵权/违法违规/事实不符，请点击【内容举报】进行投诉反馈！