c++ - libjpeg:如何获得正确的图像解压大小？

Question

我有这样一个常用的代码:

struct jpeg_decompress_struct cinfo;
jpeg_create_decompress(&cinfo);
jpeg_stdio_src(&cinfo, infile);
jpeg_read_header(&cinfo, TRUE);

cinfo.scale_num = ?;
cinfo.scale_denom = ?;

needed_width = cinfo.output_width;
needed_height = cinfo.output_height;

如何选择 scale_num 和 scale_denum 参数来将图像缩放到所需的尺寸。例如，我想将其缩小一倍。

如果我设置scale_num = 1 , scale_denom = 2 。结果是:(1802 x 1237)到(258 x 194)

文档说:

Scale the image by the fraction scale_num/scale_denom.  Default is
1/1, or no scaling.  Currently, the only supported scaling ratios
are 1/1, 1/2, 1/4, and 1/8.

但是当我设置这样的比例时，我没有得到所需的结果。

所以问题是:如何设置scale_num和scale_denom获取与所需尺寸最大相似的图像。

Answer 1

您应该调用jpeg_calc_output_dimensions(cinfo)来获取正确的output_width和output_height值。

为了计算系数scale_denom，我通常会这样做:

unsigned int intlog2(unsigned int val) {
 int targetlevel = 0;
 while (val >>= 1) ++targetlevel;
 return targetlevel;
}

// ....
// for example, 
const int width = 5184;
const int height = 3456;

unsigned int needed_width = 640;
unsigned int needed_height = 480;

unsigned int wdeg = intlog2(width / needed_width);
unsigned int hdeg = intlog2(height / needed_height);

unsigned int scale_den = std::min(1 << (std::min)(wdeg, hdeg), 8);

unsigned int result_width = width / scale_den;
unsigned int result_height = height / scale_den;