JPEG Image Rotation via DCT Coefficient Reordering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for rotating encoded JPEG images are memory intensive, requiring temporary storage of uncompressed images and multiple encoding and decoding steps, which increases memory requirements and latency.
Innovation Solution
The technique involves rotating encoded images in the DCT domain without fully decoding them to the pixel domain, by reordering and rotating the MCUs, and applying file size constraints, thereby reducing memory requirements and eliminating the need for separate coding processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional decoding and re-encoding methods are used to rotate encoded JPEG images, then the image can be rotated, but memory usage increases significantly due to temporary storage of uncompressed images
Solution Approach 1:
The image is divided into multiple 8x8 pixel blocks that are processed independently in the DCT domain. Each block's DCT coefficients are rotated separately without requiring full image decomposition, enabling localized transformation that reduces memory requirements while achieving overall image rotation.
Solution Approach 2:
The rotation operation is performed in the DCT frequency domain rather than the spatial pixel domain. By transforming the rotation operation from spatial domain to frequency domain, the patent avoids the need to decompress and store full uncompressed image data, thus reducing memory usage while achieving the rotation effect.
2Ease of operation
If conventional decoding and re-encoding methods are used to rotate encoded JPEG images, then the image can be rotated, but processing time increases due to multiple encoding and decoding steps
Solution Approach 1:
The DCT transformation is performed once during the initial encoding process, and the resulting DCT coefficients are stored. When rotation is needed, only the coefficient reordering operation is performed without repeating the full DCT transformation, significantly reducing processing time while maintaining rotation capability.
Solution Approach 2:
The patent extracts and manipulates only the essential DCT coefficient data needed for rotation, separating this critical transformation from the full encode-decode cycle. By working directly with the extracted DCT coefficients rather than reprocessing entire image data through multiple encoding/decoding stages, processing time is reduced while achieving the desired rotation effect.
3Ease of operation
If multiple encoding and decoding steps are used for image rotation, then rotation can be achieved, but computational complexity increases
Solution Approach 1:
The complex rotation operation is segmented into simple coefficient reordering steps performed on individual 8x8 blocks. Instead of applying complex transformations to entire images through multiple encode-decode cycles, the patent divides the operation into manageable block-level coefficient manipulations, reducing computational complexity while achieving accurate rotation.
Solution Approach 2:
The patent shifts the computational operation from the computationally intensive spatial domain (requiring multiple encode-decode steps) to the frequency domain where rotation can be achieved through simpler coefficient reordering. This dimensional change from spatial to frequency domain operations significantly reduces computational complexity while maintaining rotation accuracy.
Data Source
AI summary
This disclosure describes techniques for rotating an encoded image, such as an image encoded according to a JPEG standard. In one example, a method for rotating an encoded image comprising reordering minimum coded units (MCUs) of the encoded image according to a specified rotation of the encoded image, rotating image data within the MCUs according to the specified rotation, and generating a rotated version of the encoded image comprising the reordered MCUs and the rotated image data within the MCUs.