| Summary: | 碩士 === 中原大學 === 電機工程研究所 === 102 === In this thesis, we present two methods to enhance contrast for low dose x-ray images, which are Parallel Histogram Equalization (PHE) and Local Histogram Equalization (LHE). The Parallel Histogram Equalization is a very useful method to reveal X-Ray image information for abdominal, skeleton and thoracic cavity. By the way, the method’s operation time is very short. Hence, it is a great method to give consideration to image information and operation time. Another method, the Local Histogram Equalization has very large range to adjust the image contrast. But the method has a shortcoming that operation time relative to the other methods is long. The operation time we can accord with our need contrast to adjust the parameters that can reduce the operation time.
In our two proposed methods, we combine the three methods from DIP to modify the image, which were Histogram Equalization (HE), Otsu’s Threshold and Label. The Parallel Histogram Equalization uses the horizontal array and vertical array to calculate the histogram and CDF. And we do the Histogram Equalization form horizontal array’s and vertical array’s CDF. Hence, we can obtain two images, and we have to decide two image weights to combine the two images. So we can get a new image, and we use Otsu’s Threshold and Label to modify image background. Finally, we get the complete image by Parallel Histogram Equalization. Another proposed method, Local Histogram Equalization uses the mask to do Histogram Equalization. Hence, we have to decide two parameters, which are mask size and distance by neighborhood masks. The parameter one, the mask size can be used to adjust image contrast. The other parameter, the distance would affect nature of the image. And we use the Otsu’s Threshold and Label to do the Histogram Equalization for the foreground. Finally, we got newly image by Local Histogram Equalization.
The contributions of our research are as follows:
(1)Parallel Histogram Equalization can reveal image information more accurately than previous methods.
(2)Local Histogram Equalization has much better effect to control the image contrast.
(3)Our methods can make the information as clear as possible to general X-Ray image from low dose X-Ray image; hence, we hope we can decrease the hurt for our body by using the low dose X-Ray in the future.
|
|---|
