Abstract: Welds exist widely in industrial sites, and the quality inspection of welds based on X-ray flaw detection is very important to maintain the normal operation of production and life. However, high-quality X-ray flaw detection images have high bit counts. Therefore, the weld image processing system has problems such as slow reading, high power consumption and poor real-time performance. Therefore, this paper proposed a welding seam X-ray image compression method based on Walsh-Hadamard transform. Image optimization was realized based on the energy concentration property of Walsh-Hadamard transform, and key feature boundaries of weld images were preserved. The quantum particle swarm optimization algorithm was used to realize the adaptive optimization of the compression parameters of the Walsh-Hadamard transformation of the weld image. The actual production site image verification results showed that while the size of the compressed image was one-tenth of the original image, the method in this paper can preserve the weld features and image quality to the greatest extent, making the subsequent weld detection based on X-ray images more efficient.