More principle components will always give a better reconstruction of the image; it's more a measure of how many you need to get a "good" representation. Usually I think you'd look at the eigenvalues of the components to see how much variation is represented in that component. If the first eigenvalue were .9 for example, and the second .05, and the third .01, then the first three components represent 99% of the variation in the set.
Been a while since I used PCA, so that description is more colloquial; the formal answer has details that I"m leaving out, such as ensuring the components are sorted by highest eigenvalue etc...
I suggest you study the subject before using the tools.
This article can help you with your question:
Ledesma, R. D., & Valero-mora, P. (2007). Determining the Number of Factors to Retain in EFA : an easy-to- use computer program for carrying out Parallel Analysis. Practical Assessment, Research & Evaluation, 12(2). Retrieved from http://pareonline.net/getvn.asp?v=12&n=2
- Badges
- Science topic
- Similar topics
- Data Mining
- Classification