The LOT: transform coding without blocking effects

Abstract

The Lapped Orthogonal Transform (LOT) is a new tool for block transform coding with basis functions that overlap adjacent blocks [1]-[3]. The LOT can reduce the blocking effect to very low levels. In this paper, an exact derivation of an optimal LOT is presented. The optimal LOT is related to the discrete cosine transform (DCT) in such a way that a fast algorithm for a nearly optimal LOT is derived. Compared to the DCT, the fast LOT requires about 20-30 percent more computations, mostly additions. An image coding example demonstrates the effectiveness of the LOT in reducing blocking effects. Unlike earlier approaches to the reduction of blocking effects, the LOT actually leads to slightly smaller signal reconstruction errors than does the DCT.

Compresión de imágenes médicas

“…El proceso de decodificación se hace de forma directa sobre el vector de cuantificación, reemplazando los índices por su elemento correspondiente. bloqueo (discontinuidades en la reconstrucción de la señal en los límites del bloque) en la codificación de la imagen (67). Con el fin de mantener la transformación directa e inversa (la transpuesta) de las matrices, cada función base de la transformación ortogonal superpuesta debe serlo, no solo a las otras funciones del mismo bloque, sino a las de los dos bloques adyacentes (68).…”

“…Algunos trabajos (49,66,67) nos permiten concluir que las técnicas de compresión vector de cuantificación y la pirámide de Laplace hacen posible detectar y caracterizar anormalidades mucho más eficientemente en las imágenes de mamografía. Ambas técnicas deben su eficiencia de compresión a la uniformidad presente en los datos médicos.…”

Optimized nonorthogonal transforms for image compression

“…Example 2: For the case of uncorrelated, stationary blocks, (2) and (3) lead to Setting as the first block KLT vector yields 2 Throughout this paper we will assume that K(i;i) has full rank, permitting nontrivial solutions , which becomes the first stage of the block KLT 3 . The iterative algorithm described above reduces to the "power method" for the computation of eigenvectors [6].…”

“…As a result, numerous transforms like the block transforms-block Karhunen-Loeve transform (KLT), discrete cosine transform (DCT), etc. [1], lapped orthogonal transform (LOT) [2], -band perfect reconstruction filterbanks [3]-have emerged with various degrees of success in image coding applications. Clearly, all of these methods can be viewed as linear transforms and one can account for their distinctive features as structural and statistical constraints imposed on the design.…”

Reduction of Blocking Artifacts in Both Spatial Domain and Transformed Domain

“…Methods using interleaved image blocks before the encoding were also suggested [2], but they are not in conformity with the coding standards. Lowpass filtering over the block-based coded images were also proposed [3], but it may cause serious distortions when the image contains high frequency components.…”