libjpeg-turbo

libjpeg-turbo was created to provide faster JPEG encoding and decoding than the standard libjpeg library. It maintains full ABI and API compatibility with libjpeg, making it easy to replace in existing projects while gaining significant performance improvements.

libjpeg-turbo is a high-speed version of the standard JPEG image codec library, using SIMD instructions (MMX, SSE2, NEON, AltiVec) to accelerate JPEG compression and decompression.

Installation

Usage

libjpeg-turbo provides functions to read, write, and manipulate JPEG images efficiently. It supports standard JPEG features, including color spaces, sampling factors, and progressive JPEGs. Applications often use it for image viewers, editors, and servers requiring fast JPEG processing.

#include <stdio.h>
#include <jpeglib.h>
int main() {
    FILE *infile = fopen("image.jpg", "rb");
    struct jpeg_decompress_struct cinfo;
    struct jpeg_error_mgr jerr;

    cinfo.err = jpeg_std_error(&jerr);
    jpeg_create_decompress(&cinfo);
    jpeg_stdio_src(&cinfo, infile);
    jpeg_read_header(&cinfo, TRUE);
    jpeg_start_decompress(&cinfo);

    int row_stride = cinfo.output_width * cinfo.output_components;
    JSAMPARRAY buffer = (*cinfo.mem->alloc_sarray)((j_common_ptr)&cinfo, JPOOL_IMAGE, row_stride, 1);

    while (cinfo.output_scanline < cinfo.output_height) {
        jpeg_read_scanlines(&cinfo, buffer, 1);
        // Process buffer[0] here
    }

    jpeg_finish_decompress(&cinfo);
    jpeg_destroy_decompress(&cinfo);
    fclose(infile);
    return 0;
}

#include <stdio.h>
#include <jpeglib.h>
int main() {
    FILE *outfile = fopen("output.jpg", "wb");
    struct jpeg_compress_struct cinfo;
    struct jpeg_error_mgr jerr;

    cinfo.err = jpeg_std_error(&jerr);
    jpeg_create_compress(&cinfo);
    jpeg_stdio_dest(&cinfo, outfile);

    cinfo.image_width = 100;
    cinfo.image_height = 100;
    cinfo.input_components = 3;
    cinfo.in_color_space = JCS_RGB;

    jpeg_set_defaults(&cinfo);
    jpeg_start_compress(&cinfo, TRUE);

    JSAMPROW row_pointer[1];
    unsigned char row[100*3]; // Fill row with pixel data
    while(cinfo.next_scanline < cinfo.image_height) {
        row_pointer[0] = row;
        jpeg_write_scanlines(&cinfo, row_pointer, 1);
    }

    jpeg_finish_compress(&cinfo);
    jpeg_destroy_compress(&cinfo);
    fclose(outfile);
    return 0;
}

#include <turbojpeg.h>
tjhandle handle = tjInitDecompress();
int width, height, jpegSubsamp;
unsigned char *jpegBuf;
tjDecompressHeader3(handle, jpegBuf, jpegSize, &width, &height, &jpegSubsamp);
unsigned char *dstBuf = malloc(width*height*3);
tjDecompress2(handle, jpegBuf, jpegSize, dstBuf, width, 0, height, TJPF_RGB, TJFLAG_FASTDCT);
tjDestroy(handle);

tjhandle handle = tjInitCompress();
tjCompress2(handle, srcBuf, width, 0, height, TJPF_RGB, &jpegBuf, &jpegSize, TJSAMP_420, 90, TJFLAG_FASTDCT);
tjDestroy(handle);

Error Handling

Best Practices

Use the TurboJPEG API for performance-critical JPEG encoding/decoding.

Always free buffers and destroy handles to prevent memory leaks.

Leverage SIMD-optimized flags for faster computation.

Validate image dimensions and color formats before processing.

Use libjpeg-turbo in place of libjpeg where speed is essential.