An Image Enhancement Algorithm Based On Image Classification

The purpose of image enhancement is to improve the interpretability orperception of information in images for human viewers, weakening or removingunneeded information, providing better input for other automated image processingtechniques. So the algorithm design is needed to give full consideration to the purposeof image processing and the characteristics of the image itself.With the development of image processing technology, image enhancementtechniques need to be applied in more and more areas, and also asked theever-increasing demands. Common image enhancement algorithms are used forspecific application areas, and these specific areas of the image tend to have similarstatistical properties, once the image is changed, it no longer has the original statisticalproperties, the effect of the image enhancement will be no more better, so traditionalimage enhancement algorithm is often lack of universality.In order to solve the problem that common image enhancement algorithms cannot make flexible adjustment according to the characteristics of image, this paperpresented an new image enhancement algorithms based on classification.First, countthe statistical properties of image, and classify the image according to the statistics ofthe gray level distribution; according to the characteristics of the salt pepper noise andhuman visual characteristics， provide a new method of de-noising based onneighborhood extreme value, and adopt a different approach to normal pixels and thosepixels affected by noise pollution; then based on image classification, taking anadaptive image contrast enhancement can overcome the blindness of traditional imageenhancement algorithm; last, use some of the advantages of DSP and FPGA, design aUSB interface, which connected image video signal processor based on DSP andFPGA with the computer, can meet the high data transmission requirements, improvethe image and video signal processing real-time processing system, operability, andscalability.Either subjective assessment or objective assessment, the experimental resultsshow that the image enhancement method based on the classification used in this paper,is superior to the traditional median filter algorithm in the de-noising effect of the saltpepper noise, and improve the effect of the traditional histogram equalization. With theUSB interface, this paper extended the application of video and image processing system.