Bayes risk weighted vector quantization with posterior estimation for image compression and classification

Abstract

Classification and compression play important roles in communicating digital information. Their combination is useful in many applications, including the detection of abnormalities in compressed medical images. In view of the similarities of compression and low-level classification, it is not surprising that there are many similar methods for their design. Because some of these methods are useful for designing vector quantizers, it seems natural that vector quantization (VQ) is explored for the combined goal. We investigate several VQ-based algorithms that seek to minimize both the distortion of compressed images and errors in classifying their pixel blocks. These algorithms are investigated with both full search and tree-structured codes. We emphasize a nonparametric technique that minimizes both error measures simultaneously by incorporating a Bayes risk component into the distortion measure used for the design and encoding. We introduce a tree-structured posterior estimator to produce the class posterior probabilities required for the Bayes risk computation in this design. For two different image sources, we demonstrate that this system provides superior classification while maintaining compression close or superior to that of several other VQ-based designs, including Kohonen's (1992) "learning vector quantizer" and a sequential quantizer/classifier design.

Optimality of KLT for High-Rate Transform Coding of Gaussian Vector-Scale Mixtures: Application to Reconstruction, Estimation, and Classification

“…Similar joint design methodology has proven attractive in recent applications. For example, compression algorithms for medical diagnosis can be designed balancing accuracy of automatic detection of abnormality against visual signal fidelity [28,29].…”

“…Examples include denoising of natural images [23], estimation of camera motion from compressed video [24], automatic target recognition [25,26,27] and detection of abnormalities in compressed medical images [28,29]. Compression techniques can be designed to be optimal in a minimum mean squared error (MMSE) sense [30,31,32].…”

Multiresolution image classification by hierarchical modeling with two-dimensional hidden Markov models

“…On average, CART and LVQ1 perform about equally well. In [34], the Bayes VQ algorithm was used to segment the aerial images. BVQ achieves an error rate of about 21.5%, nearly the same as that of CART.…”

Can Learning Vector Quantization be an Alternative to SVM and Deep Learning? - Recent Trends and Advanced Variants of Learning Vector Quantization for Classification Learning

“…(1). The respective problem of generative (representative) models versus classification property was generally discussed for vector quantization [9,37,38,39]. In consequence, if a generative GLVQ models is strictly demanded, one has to add a respective penalty term to the cost function according to…”